($INBOX_DIR/description missing)
help / color / mirror / Atom feed[PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
57+ messages / 4 participants
[nested] [flat]
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v3 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 514 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 233 ++++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 76 +++
.../detach-partition-concurrently-2.spec | 41 ++
28 files changed, 1072 insertions(+), 171 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3927b1030d..6731c28993 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4481,6 +4481,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..1ff6214fa7 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 67144aa3c9..e4a4064511 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2544,10 +2544,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3774,7 +3776,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 0974f3e23a..6a3d033780 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1669,7 +1669,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2255,7 +2255,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index 239ac017fa..1e651ca763 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 75552c64ed..0aae01ad9f 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1104,7 +1104,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1161,7 +1161,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3788,6 +3788,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 2784fb11f9..29b12bb8a2 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -543,7 +543,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -553,7 +554,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +993,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3165,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3364,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3983,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4355,6 +4369,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4740,7 +4759,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6167,7 +6191,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6400,7 +6424,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6554,7 +6579,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6654,17 +6679,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7830,7 +7860,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8294,7 +8325,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8909,7 +8941,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9043,7 +9075,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10824,7 +10856,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11208,7 +11241,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14103,7 +14137,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14114,12 +14148,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14133,7 +14225,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14149,7 +14241,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16257,7 +16351,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16317,7 +16411,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16905,105 +16999,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17076,17 +17291,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17287,7 +17583,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17441,7 +17737,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 3b4fbdadf4..ee3fc5e39f 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 33d2c6f63d..f7ef63a1ba 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -564,6 +564,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -574,7 +575,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1054,7 +1055,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, false);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1636,7 +1637,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1702,9 +1703,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 2b4d7654cc..94dab5f5c6 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4667,6 +4667,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e2d1b987bf..8fb1d082e5 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2931,6 +2931,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index f9d0d67aa7..0fe053be96 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2123,10 +2123,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 480d168346..45be2677dc 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -651,7 +651,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2054,12 +2054,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2067,8 +2068,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2083,6 +2097,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15148,6 +15163,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15686,6 +15702,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 66c42b5898..4a7445e8f5 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..301d1c6844 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9a35147b26..9c6a1c13c5 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1275,6 +1275,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index acf8a44f30..348b4f7dd2 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 6bb0316bd9..9bcfb1d5fd 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2113,7 +2113,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2131,17 +2136,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3193,9 +3200,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3204,14 +3220,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3261,11 +3277,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 60c2f45466..95cee42c5f 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1847,6 +1848,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 71dcdf2889..2d523568c1 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..ce493e3b93
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,233 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec: EXECUTE f(1);
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2d: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2d s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2d: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2d s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2d: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index aa386ab1a2..c381fcd967 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..f035db8c5d
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,76 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1exec" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2d" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2d" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--J/dobhs11T7y2rNN--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v6 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index db29905e91..95293788d4 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4485,6 +4485,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 9abc4a1f55..1a05f2a2fe 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2555,10 +2555,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3827,7 +3829,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 4ef61b5efd..28f04a0700 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1836,7 +1836,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2486,7 +2486,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 5ab7902827..3fd0880ca0 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -430,6 +441,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -449,10 +461,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -479,6 +502,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 8bc652ecd3..82670bac21 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -410,7 +410,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1123,7 +1123,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1180,7 +1180,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -4073,6 +4073,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index b990063d38..91ef4760f6 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3266,7 +3274,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3465,7 +3473,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -4084,7 +4092,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4475,6 +4490,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4860,7 +4880,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6292,7 +6317,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6525,7 +6550,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6679,7 +6705,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6779,17 +6805,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7388,7 +7419,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7971,7 +8002,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8435,7 +8467,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -9050,7 +9083,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9184,7 +9217,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10968,7 +11001,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11352,7 +11386,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14173,7 +14208,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14184,12 +14219,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14203,7 +14296,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14219,7 +14312,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16327,7 +16422,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16387,7 +16482,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16976,105 +17071,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17147,22 +17367,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17358,7 +17706,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17512,7 +17860,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 8908847c6c..fb2cdbad6d 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1119,7 +1119,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1141,7 +1141,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index b8da4c5967..619aaffae4 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -569,6 +569,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation partrel;
int firstVarno = mtstate->resultRelInfo[0].ri_RangeTableIndex;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -579,7 +580,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1065,9 +1066,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1645,9 +1658,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1713,9 +1727,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 65bbc18ecb..1cd2931a74 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4713,6 +4713,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index c2d73626fc..250a23b347 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2972,6 +2972,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 177e6e336a..87b21dcbdc 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2136,10 +2136,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dd72a9fc3c..6d2dc11b23 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -646,7 +646,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2096,12 +2096,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2109,8 +2110,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2125,6 +2139,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15337,6 +15352,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15877,6 +15893,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index e5f3482d52..2b2b1dc1ad 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2798,7 +2798,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -3990,7 +3990,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4190,7 +4190,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 05bb698cf4..729274b330 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1236,6 +1236,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 20af5a92b4..a8d64014f0 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 2b71cad9a2..155596907c 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid BKI_LOOKUP(pg_class);
Oid inhparent BKI_LOOKUP(pg_class);
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 236832a2ca..f97105ed3a 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1896,6 +1897,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 28083aaac9..7da248a385 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -165,6 +165,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 2c8b881a09..901eccec25 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 5d6b79e66e..69928dc540 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--LZvS9be/3tNcYl/X--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v8 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 31 +-
src/backend/catalog/pg_inherits.c | 91 ++-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 558 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/clauses.c | 8 +-
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 16 +-
src/bin/psql/describe.c | 42 +-
src/include/catalog/pg_inherits.h | 8 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 237 ++++++++
.../detach-partition-concurrently-2.out | 66 +++
.../detach-partition-concurrently-3.out | 211 +++++++
src/test/isolation/isolation_schedule | 3 +
.../detach-partition-concurrently-1.spec | 69 +++
.../detach-partition-concurrently-2.spec | 41 ++
.../detach-partition-concurrently-3.spec | 50 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 175 ++++++
.../specs/partition-removal-1.spec | 58 ++
src/test/regress/expected/alter_table.out | 29 +
src/test/regress/sql/alter_table.sql | 20 +
37 files changed, 1735 insertions(+), 165 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-3.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-3.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 5c9f4af1d5..0cb846f408 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4485,6 +4485,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetachpending</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 9abc4a1f55..1a05f2a2fe 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2555,10 +2555,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3827,7 +3829,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 4ef61b5efd..28f04a0700 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1836,7 +1836,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2486,7 +2486,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..797a7384ad 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -68,15 +74,20 @@ get_partition_parent(Oid relid)
* get_partition_parent_worker
* Scan the pg_inherits relation to return the OID of the parent of the
* given relation
+ *
+ * If the partition is being detached, InvalidOid is returned, and also
+ * *detached is set true.
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +104,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetachpending)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +152,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 5ab7902827..f62720c647 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetachpending &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -430,6 +441,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetachpending - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -449,10 +461,17 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * expect_detach_pending is the expected state of the inhdetachpending flag.
+ * If the catalog row does not match that state, an error is raised.
+ *
+ * childname is the partition name, if a table; pass NULL for regular
+ * inheritance or when working with other relation kinds.
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool expect_detach_pending,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -479,6 +498,29 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool detach_pending;
+
+ detach_pending =
+ ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetachpending;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (detach_pending && !expect_detach_pending)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname ? childname : "unknown relation"),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!detach_pending && expect_detach_pending)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname ? childname : "unknown relation"),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
@@ -490,3 +532,46 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
return found;
}
+
+/*
+ * Return whether the pg_inherits tuple for a partition has the "detach
+ * pending" flag set.
+ */
+bool
+PartitionHasPendingDetach(Oid partoid)
+{
+ Relation catalogRelation;
+ ScanKeyData key;
+ SysScanDesc scan;
+ HeapTuple inheritsTuple;
+
+ /* We don't have a good way to verify it is in fact a partition */
+
+ /*
+ * Find the pg_inherits entry by inhrelid. (There should only be one.)
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(partoid));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ bool detached;
+
+ detached =
+ ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetachpending;
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ return detached;
+ }
+
+ elog(ERROR, "relation %u is not a partition", partoid);
+ return false; /* keep compiler quiet */
+}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 8bc652ecd3..676edee233 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -410,7 +410,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1123,7 +1123,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1180,7 +1180,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -4073,6 +4073,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetachpending - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 8e753f1efd..81e866c1eb 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -550,8 +551,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -986,7 +993,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3265,7 +3273,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3464,7 +3472,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -4083,7 +4091,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4167,6 +4182,19 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* Find or create work queue entry for this table */
tab = ATGetQueueEntry(wqueue, rel);
+ /*
+ * Disallow any ALTER TABLE other than ALTER TABLE DETACH FINALIZE on
+ * partitions that are pending detach.
+ */
+ if (rel->rd_rel->relispartition &&
+ cmd->subtype != AT_DetachPartitionFinalize &&
+ PartitionHasPendingDetach(RelationGetRelid(rel)))
+ ereport(ERROR,
+ errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot alter partition \"%s\" with an incomplete detach",
+ RelationGetRelationName(rel)),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation."));
+
/*
* Copy the original subcommand for each table. This avoids conflicts
* when different child tables need to make different parse
@@ -4474,6 +4502,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4859,7 +4892,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6291,7 +6329,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6524,7 +6562,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6678,7 +6717,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7387,7 +7426,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7970,7 +8009,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8434,7 +8474,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -9049,7 +9090,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9183,7 +9224,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10967,7 +11008,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11351,7 +11393,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14172,7 +14215,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14183,12 +14226,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetachpending = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14202,7 +14303,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14218,7 +14319,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16303,7 +16406,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16363,7 +16466,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16952,105 +17055,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17123,22 +17351,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17334,7 +17690,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17488,7 +17844,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 4e4e05844c..7383d5994e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1119,7 +1119,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1141,7 +1141,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index b8da4c5967..619aaffae4 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -569,6 +569,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation partrel;
int firstVarno = mtstate->resultRelInfo[0].ri_RangeTableIndex;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -579,7 +580,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1065,9 +1066,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1645,9 +1658,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1713,9 +1727,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index da91cbd2b1..850e50f1a2 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4735,6 +4735,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index c2d73626fc..250a23b347 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2972,6 +2972,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/clauses.c b/src/backend/optimizer/util/clauses.c
index ade66a541c..6ff5b03af2 100644
--- a/src/backend/optimizer/util/clauses.c
+++ b/src/backend/optimizer/util/clauses.c
@@ -1184,11 +1184,15 @@ target_rel_partitions_max_parallel_hazard(Relation rel,
/* Recursively check each partition ... */
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already. Must
+ * match args used in set_relation_partition_info.
+ */
glob = context->planner_global;
if (glob->partition_directory == NULL)
glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
pdesc = PartitionDirectoryLookup(glob->partition_directory, rel);
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index c5947fa418..9ec7d7f518 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2142,10 +2142,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 652be0b96d..b26d0a3d0b 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -646,7 +646,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2096,12 +2096,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2109,8 +2110,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2125,6 +2139,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15348,6 +15363,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15888,6 +15904,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index e5f3482d52..2b2b1dc1ad 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2798,7 +2798,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -3990,7 +3990,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4190,7 +4190,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 05bb698cf4..729274b330 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1236,6 +1236,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..0fe4f55b04 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,20 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached. The constraint is NIL in that case, and let's cache that.
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ rel->rd_partcheck = NIL;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 20af5a92b4..eb3e79adf7 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetachpending" :
+ "false as inhdetachpending");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACH PENDING" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,20 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
"SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ " inhdetachpending,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ " false AS inhdetachpending,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3226,16 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ " false AS inhdetachpending, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ " false AS inhdetachpending, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3285,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACH PENDING)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 2b71cad9a2..6d07e1b302 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid BKI_LOOKUP(pg_class);
Oid inhparent BKI_LOOKUP(pg_class);
int32 inhseqno;
+ bool inhdetachpending;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,8 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
+extern bool PartitionHasPendingDetach(Oid partoid);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 236832a2ca..f97105ed3a 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1896,6 +1897,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 28083aaac9..7da248a385 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -165,6 +165,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 2c8b881a09..901eccec25 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..9b813b9a75
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,237 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/expected/detach-partition-concurrently-3.out b/src/test/isolation/expected/detach-partition-concurrently-3.out
new file mode 100644
index 0000000000..394ec2b40d
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-3.out
@@ -0,0 +1,211 @@
+Parsed test spec with 2 sessions
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1alter
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1alter: ALTER TABLE d3_listp1 ALTER a DROP NOT NULL;
+ERROR: cannot alter partition "d3_listp1" with an incomplete detach
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1insert
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1insert: INSERT INTO d3_listp VALUES (1);
+ERROR: no partition of relation "d3_listp" found for row
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1drop s1spart
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1drop: DROP TABLE d3_listp;
+step s1spart: SELECT * FROM d3_listp1;
+ERROR: relation "d3_listp1" does not exist
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1trunc s1spart
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1trunc: TRUNCATE TABLE d3_listp;
+step s1spart: SELECT * FROM d3_listp1;
+a
+
+1
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1b s1s s2detachfinal s1c
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE;
+step s1c: COMMIT;
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1b s1spart s2detachfinal s1c
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1b: BEGIN;
+step s1spart: SELECT * FROM d3_listp1;
+a
+
+1
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE; <waiting ...>
+step s1c: COMMIT;
+step s2detachfinal: <... completed>
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1b s1spart s2detachfinal s1c
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1b: BEGIN;
+step s1spart: SELECT * FROM d3_listp1;
+a
+
+1
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE; <waiting ...>
+step s1c: COMMIT;
+step s2detachfinal: <... completed>
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s2begin s2detachfinal s2commit
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s2begin: BEGIN;
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE;
+step s2commit: COMMIT;
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s2begin s2detachfinal s1spart s2commit
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s2begin: BEGIN;
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE;
+step s1spart: SELECT * FROM d3_listp1; <waiting ...>
+step s2commit: COMMIT;
+step s1spart: <... completed>
+a
+
+1
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s2begin s2detachfinal s1insertpart s2commit
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s2begin: BEGIN;
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE;
+step s1insertpart: INSERT INTO d3_listp1 VALUES (1); <waiting ...>
+step s2commit: COMMIT;
+step s1insertpart: <... completed>
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 5d6b79e66e..25cd7d0a13 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,9 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
+test: detach-partition-concurrently-3
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..6ca0adac1d
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,69 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-3.spec b/src/test/isolation/specs/detach-partition-concurrently-3.spec
new file mode 100644
index 0000000000..5f5d650791
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-3.spec
@@ -0,0 +1,50 @@
+# Try various things to happen to a partition with an incomplete detach
+
+setup
+{
+ CREATE TABLE d3_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d3_listp1 PARTITION OF d3_listp FOR VALUES IN (1);
+ CREATE TABLE d3_pid (pid int);
+ INSERT INTO d3_listp VALUES (1);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d3_listp, d3_listp1, d3_pid;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d3_listp; }
+step "s1spart" { SELECT * FROM d3_listp1; }
+step "s1cancel" { SELECT pg_cancel_backend(pid) FROM d3_pid; }
+step "s1c" { COMMIT; }
+step "s1alter" { ALTER TABLE d3_listp1 ALTER a DROP NOT NULL; }
+step "s1insert" { INSERT INTO d3_listp VALUES (1); }
+step "s1insertpart" { INSERT INTO d3_listp1 VALUES (1); }
+step "s1drop" { DROP TABLE d3_listp; }
+step "s1trunc" { TRUNCATE TABLE d3_listp; }
+
+session "s2"
+step "s2begin" { BEGIN; }
+step "s2snitch" { INSERT INTO d3_pid SELECT pg_backend_pid(); }
+step "s2detach" { ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; }
+step "s2detachfinal" { ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE; }
+step "s2commit" { COMMIT; }
+
+# Try various things while the partition is in "being detached" state, with
+# no session waiting.
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1alter"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1insert"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1drop" "s1spart"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1trunc" "s1spart"
+
+# partially detach, then read from partition and try to complete the detach
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1b" "s1s" "s2detachfinal" "s1c"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1b" "s1spart" "s2detachfinal" "s1c"
+
+# DETACH FINALIZE in a transaction block. No insert/select on the partition
+# is allowed concurrently with that.
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1b" "s1spart" "s2detachfinal" "s1c"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s2begin" "s2detachfinal" "s2commit"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s2begin" "s2detachfinal" "s1spart" "s2commit"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s2begin" "s2detachfinal" "s1insertpart" "s2commit"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..427f41c9aa
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,175 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3check s3unlock s3check s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3check: SELECT * FROM partrem;
+a b
+
+1 ABC
+3 DEF
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s3check: SELECT * FROM partrem;
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3check s3unlock s3check s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3check: SELECT * FROM partrem;
+a b
+
+1 ABC
+3 DEF
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s3check: SELECT * FROM partrem;
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2) AND a <> 1; <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2) AND a <> 1; <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1prepare s2remp s1execprep s3unlock s1check s1c s1check s1dealloc
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prepare: PREPARE ins AS INSERT INTO partrem VALUES ($1, 'GHI');
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s1execprep: EXECUTE ins(2); <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1execprep: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+2 GHI
+step s1c: COMMIT;
+step s2remp: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+step s1dealloc: DEALLOCATE ins;
+
+starting permutation: s1brr s1prepare s2remp s3lock s1execprep s3unlock s1check s1c s1check s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prepare: PREPARE ins AS INSERT INTO partrem VALUES ($1, 'GHI');
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1execprep: EXECUTE ins(2); <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1execprep: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+2 GHI
+step s1c: COMMIT;
+step s2remp: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+step s1dealloc: DEALLOCATE ins;
+
+starting permutation: s1brr s1check s3lock s2remp s1prepare s1execprep s3unlock s1check s1c s1check s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s1prepare: PREPARE ins AS INSERT INTO partrem VALUES ($1, 'GHI');
+step s1execprep: EXECUTE ins(2); <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1execprep: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+2 GHI
+step s1c: COMMIT;
+step s2remp: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+step s1dealloc: DEALLOCATE ins;
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..5ee2750129
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,58 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2) AND a <> 1; }
+step "s1prepare" { PREPARE ins AS INSERT INTO partrem VALUES ($1, 'GHI'); }
+step "s1execprep" { EXECUTE ins(2); }
+step "s1check" { SELECT * FROM partrem WHERE b = 'GHI'; }
+step "s1c" { COMMIT; }
+step "s1dealloc" { DEALLOCATE ins; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+step "s3check" { SELECT * FROM partrem; }
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway (so its row doesn't
+# show up in s3's result), but we expect its row to still appear in the
+# result for s1.
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3check" "s3unlock" "s3check" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3check" "s3unlock" "s3check" "s1c"
+
+# In this case we're testing that after pruning partrem2 at runtime, the
+# query still works correctly.
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
+
+# In this case we test that an insert that's prepared in repeatable read
+# mode still works after detaching.
+permutation "s3lock" "s1brr" "s1prepare" "s2remp" "s1execprep" "s3unlock" "s1check" "s1c" "s1check" "s1dealloc"
+permutation "s1brr" "s1prepare" "s2remp" "s3lock" "s1execprep" "s3unlock" "s1check" "s1c" "s1check" "s1dealloc"
+permutation "s1brr" "s1check" "s3lock" "s2remp" "s1prepare" "s1execprep" "s3unlock" "s1check" "s1c" "s1check" "s1dealloc"
diff --git a/src/test/regress/expected/alter_table.out b/src/test/regress/expected/alter_table.out
index bb3f873f22..ec14b060a6 100644
--- a/src/test/regress/expected/alter_table.out
+++ b/src/test/regress/expected/alter_table.out
@@ -4163,6 +4163,35 @@ SELECT * from part_rp;
(0 rows)
DROP TABLE part_rp;
+-- concurrent detach
+CREATE TABLE range_parted2 (
+ a int
+) PARTITION BY RANGE(a);
+CREATE TABLE part_rp PARTITION OF range_parted2 FOR VALUES FROM (0) to (100);
+BEGIN;
+-- doesn't work in a partition block
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+ERROR: ALTER TABLE ... DETACH CONCURRENTLY cannot run inside a transaction block
+COMMIT;
+CREATE TABLE part_rpd PARTITION OF range_parted2 DEFAULT;
+-- doesn't work if there's a default partition
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+ERROR: cannot detach partitions concurrently when a default partition exists
+-- doesn't work for the default partition
+ALTER TABLE range_parted2 DETACH PARTITION part_rpd CONCURRENTLY;
+ERROR: cannot detach partitions concurrently when a default partition exists
+DROP TABLE part_rpd;
+-- works fine
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+\d+ range_parted2
+ Partitioned table "public.range_parted2"
+ Column | Type | Collation | Nullable | Default | Storage | Stats target | Description
+--------+---------+-----------+----------+---------+---------+--------------+-------------
+ a | integer | | | | plain | |
+Partition key: RANGE (a)
+Number of partitions: 0
+
+DROP TABLE range_parted2;
-- Check ALTER TABLE commands for partitioned tables and partitions
-- cannot add/drop column to/from *only* the parent
ALTER TABLE ONLY list_parted2 ADD COLUMN c int;
diff --git a/src/test/regress/sql/alter_table.sql b/src/test/regress/sql/alter_table.sql
index 4cc55d8525..7a9c9252dc 100644
--- a/src/test/regress/sql/alter_table.sql
+++ b/src/test/regress/sql/alter_table.sql
@@ -2678,6 +2678,26 @@ DROP TABLE range_parted2;
SELECT * from part_rp;
DROP TABLE part_rp;
+-- concurrent detach
+CREATE TABLE range_parted2 (
+ a int
+) PARTITION BY RANGE(a);
+CREATE TABLE part_rp PARTITION OF range_parted2 FOR VALUES FROM (0) to (100);
+BEGIN;
+-- doesn't work in a partition block
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+COMMIT;
+CREATE TABLE part_rpd PARTITION OF range_parted2 DEFAULT;
+-- doesn't work if there's a default partition
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+-- doesn't work for the default partition
+ALTER TABLE range_parted2 DETACH PARTITION part_rpd CONCURRENTLY;
+DROP TABLE part_rpd;
+-- works fine
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+\d+ range_parted2
+DROP TABLE range_parted2;
+
-- Check ALTER TABLE commands for partitioned tables and partitions
-- cannot add/drop column to/from *only* the parent
--
2.20.1
--XsQoSWH+UP9D9v3l--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v8 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 31 +-
src/backend/catalog/pg_inherits.c | 91 ++-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 558 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/clauses.c | 8 +-
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 16 +-
src/bin/psql/describe.c | 42 +-
src/include/catalog/pg_inherits.h | 8 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 237 ++++++++
.../detach-partition-concurrently-2.out | 66 +++
.../detach-partition-concurrently-3.out | 211 +++++++
src/test/isolation/isolation_schedule | 3 +
.../detach-partition-concurrently-1.spec | 69 +++
.../detach-partition-concurrently-2.spec | 41 ++
.../detach-partition-concurrently-3.spec | 50 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 175 ++++++
.../specs/partition-removal-1.spec | 58 ++
src/test/regress/expected/alter_table.out | 29 +
src/test/regress/sql/alter_table.sql | 20 +
37 files changed, 1735 insertions(+), 165 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-3.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-3.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 5c9f4af1d5..0cb846f408 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4485,6 +4485,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetachpending</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 9abc4a1f55..1a05f2a2fe 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2555,10 +2555,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3827,7 +3829,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 4ef61b5efd..28f04a0700 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1836,7 +1836,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2486,7 +2486,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..797a7384ad 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -68,15 +74,20 @@ get_partition_parent(Oid relid)
* get_partition_parent_worker
* Scan the pg_inherits relation to return the OID of the parent of the
* given relation
+ *
+ * If the partition is being detached, InvalidOid is returned, and also
+ * *detached is set true.
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +104,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetachpending)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +152,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 5ab7902827..f62720c647 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetachpending &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -430,6 +441,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetachpending - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -449,10 +461,17 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * expect_detach_pending is the expected state of the inhdetachpending flag.
+ * If the catalog row does not match that state, an error is raised.
+ *
+ * childname is the partition name, if a table; pass NULL for regular
+ * inheritance or when working with other relation kinds.
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool expect_detach_pending,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -479,6 +498,29 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool detach_pending;
+
+ detach_pending =
+ ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetachpending;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (detach_pending && !expect_detach_pending)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname ? childname : "unknown relation"),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!detach_pending && expect_detach_pending)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname ? childname : "unknown relation"),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
@@ -490,3 +532,46 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
return found;
}
+
+/*
+ * Return whether the pg_inherits tuple for a partition has the "detach
+ * pending" flag set.
+ */
+bool
+PartitionHasPendingDetach(Oid partoid)
+{
+ Relation catalogRelation;
+ ScanKeyData key;
+ SysScanDesc scan;
+ HeapTuple inheritsTuple;
+
+ /* We don't have a good way to verify it is in fact a partition */
+
+ /*
+ * Find the pg_inherits entry by inhrelid. (There should only be one.)
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(partoid));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ bool detached;
+
+ detached =
+ ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetachpending;
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ return detached;
+ }
+
+ elog(ERROR, "relation %u is not a partition", partoid);
+ return false; /* keep compiler quiet */
+}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 8bc652ecd3..676edee233 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -410,7 +410,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1123,7 +1123,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1180,7 +1180,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -4073,6 +4073,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetachpending - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 8e753f1efd..81e866c1eb 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -550,8 +551,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -986,7 +993,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3265,7 +3273,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3464,7 +3472,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -4083,7 +4091,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4167,6 +4182,19 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* Find or create work queue entry for this table */
tab = ATGetQueueEntry(wqueue, rel);
+ /*
+ * Disallow any ALTER TABLE other than ALTER TABLE DETACH FINALIZE on
+ * partitions that are pending detach.
+ */
+ if (rel->rd_rel->relispartition &&
+ cmd->subtype != AT_DetachPartitionFinalize &&
+ PartitionHasPendingDetach(RelationGetRelid(rel)))
+ ereport(ERROR,
+ errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot alter partition \"%s\" with an incomplete detach",
+ RelationGetRelationName(rel)),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation."));
+
/*
* Copy the original subcommand for each table. This avoids conflicts
* when different child tables need to make different parse
@@ -4474,6 +4502,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4859,7 +4892,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6291,7 +6329,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6524,7 +6562,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6678,7 +6717,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7387,7 +7426,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7970,7 +8009,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8434,7 +8474,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -9049,7 +9090,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9183,7 +9224,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10967,7 +11008,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11351,7 +11393,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14172,7 +14215,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14183,12 +14226,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetachpending = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14202,7 +14303,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14218,7 +14319,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16303,7 +16406,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16363,7 +16466,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16952,105 +17055,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17123,22 +17351,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17334,7 +17690,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17488,7 +17844,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 4e4e05844c..7383d5994e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1119,7 +1119,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1141,7 +1141,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index b8da4c5967..619aaffae4 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -569,6 +569,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation partrel;
int firstVarno = mtstate->resultRelInfo[0].ri_RangeTableIndex;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -579,7 +580,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1065,9 +1066,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1645,9 +1658,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1713,9 +1727,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index da91cbd2b1..850e50f1a2 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4735,6 +4735,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index c2d73626fc..250a23b347 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2972,6 +2972,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/clauses.c b/src/backend/optimizer/util/clauses.c
index ade66a541c..6ff5b03af2 100644
--- a/src/backend/optimizer/util/clauses.c
+++ b/src/backend/optimizer/util/clauses.c
@@ -1184,11 +1184,15 @@ target_rel_partitions_max_parallel_hazard(Relation rel,
/* Recursively check each partition ... */
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already. Must
+ * match args used in set_relation_partition_info.
+ */
glob = context->planner_global;
if (glob->partition_directory == NULL)
glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
pdesc = PartitionDirectoryLookup(glob->partition_directory, rel);
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index c5947fa418..9ec7d7f518 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2142,10 +2142,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 652be0b96d..b26d0a3d0b 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -646,7 +646,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2096,12 +2096,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2109,8 +2110,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2125,6 +2139,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15348,6 +15363,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15888,6 +15904,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index e5f3482d52..2b2b1dc1ad 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2798,7 +2798,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -3990,7 +3990,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4190,7 +4190,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 05bb698cf4..729274b330 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1236,6 +1236,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..0fe4f55b04 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,20 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached. The constraint is NIL in that case, and let's cache that.
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ rel->rd_partcheck = NIL;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 20af5a92b4..eb3e79adf7 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetachpending" :
+ "false as inhdetachpending");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACH PENDING" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,20 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
"SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ " inhdetachpending,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ " false AS inhdetachpending,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3226,16 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ " false AS inhdetachpending, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ " false AS inhdetachpending, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3285,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACH PENDING)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 2b71cad9a2..6d07e1b302 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid BKI_LOOKUP(pg_class);
Oid inhparent BKI_LOOKUP(pg_class);
int32 inhseqno;
+ bool inhdetachpending;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,8 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
+extern bool PartitionHasPendingDetach(Oid partoid);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 236832a2ca..f97105ed3a 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1896,6 +1897,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 28083aaac9..7da248a385 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -165,6 +165,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 2c8b881a09..901eccec25 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..9b813b9a75
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,237 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/expected/detach-partition-concurrently-3.out b/src/test/isolation/expected/detach-partition-concurrently-3.out
new file mode 100644
index 0000000000..394ec2b40d
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-3.out
@@ -0,0 +1,211 @@
+Parsed test spec with 2 sessions
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1alter
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1alter: ALTER TABLE d3_listp1 ALTER a DROP NOT NULL;
+ERROR: cannot alter partition "d3_listp1" with an incomplete detach
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1insert
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1insert: INSERT INTO d3_listp VALUES (1);
+ERROR: no partition of relation "d3_listp" found for row
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1drop s1spart
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1drop: DROP TABLE d3_listp;
+step s1spart: SELECT * FROM d3_listp1;
+ERROR: relation "d3_listp1" does not exist
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1trunc s1spart
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1trunc: TRUNCATE TABLE d3_listp;
+step s1spart: SELECT * FROM d3_listp1;
+a
+
+1
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1b s1s s2detachfinal s1c
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE;
+step s1c: COMMIT;
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1b s1spart s2detachfinal s1c
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1b: BEGIN;
+step s1spart: SELECT * FROM d3_listp1;
+a
+
+1
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE; <waiting ...>
+step s1c: COMMIT;
+step s2detachfinal: <... completed>
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1b s1spart s2detachfinal s1c
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1b: BEGIN;
+step s1spart: SELECT * FROM d3_listp1;
+a
+
+1
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE; <waiting ...>
+step s1c: COMMIT;
+step s2detachfinal: <... completed>
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s2begin s2detachfinal s2commit
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s2begin: BEGIN;
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE;
+step s2commit: COMMIT;
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s2begin s2detachfinal s1spart s2commit
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s2begin: BEGIN;
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE;
+step s1spart: SELECT * FROM d3_listp1; <waiting ...>
+step s2commit: COMMIT;
+step s1spart: <... completed>
+a
+
+1
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s2begin s2detachfinal s1insertpart s2commit
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s2begin: BEGIN;
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE;
+step s1insertpart: INSERT INTO d3_listp1 VALUES (1); <waiting ...>
+step s2commit: COMMIT;
+step s1insertpart: <... completed>
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 5d6b79e66e..25cd7d0a13 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,9 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
+test: detach-partition-concurrently-3
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..6ca0adac1d
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,69 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-3.spec b/src/test/isolation/specs/detach-partition-concurrently-3.spec
new file mode 100644
index 0000000000..5f5d650791
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-3.spec
@@ -0,0 +1,50 @@
+# Try various things to happen to a partition with an incomplete detach
+
+setup
+{
+ CREATE TABLE d3_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d3_listp1 PARTITION OF d3_listp FOR VALUES IN (1);
+ CREATE TABLE d3_pid (pid int);
+ INSERT INTO d3_listp VALUES (1);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d3_listp, d3_listp1, d3_pid;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d3_listp; }
+step "s1spart" { SELECT * FROM d3_listp1; }
+step "s1cancel" { SELECT pg_cancel_backend(pid) FROM d3_pid; }
+step "s1c" { COMMIT; }
+step "s1alter" { ALTER TABLE d3_listp1 ALTER a DROP NOT NULL; }
+step "s1insert" { INSERT INTO d3_listp VALUES (1); }
+step "s1insertpart" { INSERT INTO d3_listp1 VALUES (1); }
+step "s1drop" { DROP TABLE d3_listp; }
+step "s1trunc" { TRUNCATE TABLE d3_listp; }
+
+session "s2"
+step "s2begin" { BEGIN; }
+step "s2snitch" { INSERT INTO d3_pid SELECT pg_backend_pid(); }
+step "s2detach" { ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; }
+step "s2detachfinal" { ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE; }
+step "s2commit" { COMMIT; }
+
+# Try various things while the partition is in "being detached" state, with
+# no session waiting.
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1alter"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1insert"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1drop" "s1spart"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1trunc" "s1spart"
+
+# partially detach, then read from partition and try to complete the detach
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1b" "s1s" "s2detachfinal" "s1c"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1b" "s1spart" "s2detachfinal" "s1c"
+
+# DETACH FINALIZE in a transaction block. No insert/select on the partition
+# is allowed concurrently with that.
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1b" "s1spart" "s2detachfinal" "s1c"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s2begin" "s2detachfinal" "s2commit"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s2begin" "s2detachfinal" "s1spart" "s2commit"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s2begin" "s2detachfinal" "s1insertpart" "s2commit"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..427f41c9aa
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,175 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3check s3unlock s3check s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3check: SELECT * FROM partrem;
+a b
+
+1 ABC
+3 DEF
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s3check: SELECT * FROM partrem;
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3check s3unlock s3check s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3check: SELECT * FROM partrem;
+a b
+
+1 ABC
+3 DEF
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s3check: SELECT * FROM partrem;
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2) AND a <> 1; <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2) AND a <> 1; <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1prepare s2remp s1execprep s3unlock s1check s1c s1check s1dealloc
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prepare: PREPARE ins AS INSERT INTO partrem VALUES ($1, 'GHI');
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s1execprep: EXECUTE ins(2); <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1execprep: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+2 GHI
+step s1c: COMMIT;
+step s2remp: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+step s1dealloc: DEALLOCATE ins;
+
+starting permutation: s1brr s1prepare s2remp s3lock s1execprep s3unlock s1check s1c s1check s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prepare: PREPARE ins AS INSERT INTO partrem VALUES ($1, 'GHI');
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1execprep: EXECUTE ins(2); <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1execprep: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+2 GHI
+step s1c: COMMIT;
+step s2remp: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+step s1dealloc: DEALLOCATE ins;
+
+starting permutation: s1brr s1check s3lock s2remp s1prepare s1execprep s3unlock s1check s1c s1check s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s1prepare: PREPARE ins AS INSERT INTO partrem VALUES ($1, 'GHI');
+step s1execprep: EXECUTE ins(2); <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1execprep: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+2 GHI
+step s1c: COMMIT;
+step s2remp: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+step s1dealloc: DEALLOCATE ins;
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..5ee2750129
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,58 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2) AND a <> 1; }
+step "s1prepare" { PREPARE ins AS INSERT INTO partrem VALUES ($1, 'GHI'); }
+step "s1execprep" { EXECUTE ins(2); }
+step "s1check" { SELECT * FROM partrem WHERE b = 'GHI'; }
+step "s1c" { COMMIT; }
+step "s1dealloc" { DEALLOCATE ins; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+step "s3check" { SELECT * FROM partrem; }
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway (so its row doesn't
+# show up in s3's result), but we expect its row to still appear in the
+# result for s1.
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3check" "s3unlock" "s3check" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3check" "s3unlock" "s3check" "s1c"
+
+# In this case we're testing that after pruning partrem2 at runtime, the
+# query still works correctly.
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
+
+# In this case we test that an insert that's prepared in repeatable read
+# mode still works after detaching.
+permutation "s3lock" "s1brr" "s1prepare" "s2remp" "s1execprep" "s3unlock" "s1check" "s1c" "s1check" "s1dealloc"
+permutation "s1brr" "s1prepare" "s2remp" "s3lock" "s1execprep" "s3unlock" "s1check" "s1c" "s1check" "s1dealloc"
+permutation "s1brr" "s1check" "s3lock" "s2remp" "s1prepare" "s1execprep" "s3unlock" "s1check" "s1c" "s1check" "s1dealloc"
diff --git a/src/test/regress/expected/alter_table.out b/src/test/regress/expected/alter_table.out
index bb3f873f22..ec14b060a6 100644
--- a/src/test/regress/expected/alter_table.out
+++ b/src/test/regress/expected/alter_table.out
@@ -4163,6 +4163,35 @@ SELECT * from part_rp;
(0 rows)
DROP TABLE part_rp;
+-- concurrent detach
+CREATE TABLE range_parted2 (
+ a int
+) PARTITION BY RANGE(a);
+CREATE TABLE part_rp PARTITION OF range_parted2 FOR VALUES FROM (0) to (100);
+BEGIN;
+-- doesn't work in a partition block
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+ERROR: ALTER TABLE ... DETACH CONCURRENTLY cannot run inside a transaction block
+COMMIT;
+CREATE TABLE part_rpd PARTITION OF range_parted2 DEFAULT;
+-- doesn't work if there's a default partition
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+ERROR: cannot detach partitions concurrently when a default partition exists
+-- doesn't work for the default partition
+ALTER TABLE range_parted2 DETACH PARTITION part_rpd CONCURRENTLY;
+ERROR: cannot detach partitions concurrently when a default partition exists
+DROP TABLE part_rpd;
+-- works fine
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+\d+ range_parted2
+ Partitioned table "public.range_parted2"
+ Column | Type | Collation | Nullable | Default | Storage | Stats target | Description
+--------+---------+-----------+----------+---------+---------+--------------+-------------
+ a | integer | | | | plain | |
+Partition key: RANGE (a)
+Number of partitions: 0
+
+DROP TABLE range_parted2;
-- Check ALTER TABLE commands for partitioned tables and partitions
-- cannot add/drop column to/from *only* the parent
ALTER TABLE ONLY list_parted2 ADD COLUMN c int;
diff --git a/src/test/regress/sql/alter_table.sql b/src/test/regress/sql/alter_table.sql
index 4cc55d8525..7a9c9252dc 100644
--- a/src/test/regress/sql/alter_table.sql
+++ b/src/test/regress/sql/alter_table.sql
@@ -2678,6 +2678,26 @@ DROP TABLE range_parted2;
SELECT * from part_rp;
DROP TABLE part_rp;
+-- concurrent detach
+CREATE TABLE range_parted2 (
+ a int
+) PARTITION BY RANGE(a);
+CREATE TABLE part_rp PARTITION OF range_parted2 FOR VALUES FROM (0) to (100);
+BEGIN;
+-- doesn't work in a partition block
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+COMMIT;
+CREATE TABLE part_rpd PARTITION OF range_parted2 DEFAULT;
+-- doesn't work if there's a default partition
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+-- doesn't work for the default partition
+ALTER TABLE range_parted2 DETACH PARTITION part_rpd CONCURRENTLY;
+DROP TABLE part_rpd;
+-- works fine
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+\d+ range_parted2
+DROP TABLE range_parted2;
+
-- Check ALTER TABLE commands for partitioned tables and partitions
-- cannot add/drop column to/from *only* the parent
--
2.20.1
--XsQoSWH+UP9D9v3l--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v4] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 587 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 +++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
29 files changed, 1234 insertions(+), 178 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 5fb9dca425..b628ffaa4b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..1ff6214fa7 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 4cd7d76938..38728f268e 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2544,10 +2544,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3816,7 +3818,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index b88b4a1f12..95d1370cfe 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2424,7 +2424,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index 239ac017fa..1e651ca763 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 75552c64ed..798e9c76d0 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -398,7 +398,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1104,7 +1104,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1161,7 +1161,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3788,6 +3788,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 4bb33ee03c..7d6aa91f78 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -158,6 +158,8 @@ typedef struct AlteredTableInfo
Oid relid; /* Relation to work on */
char relkind; /* Its relkind */
TupleDesc oldDesc; /* Pre-modification tuple descriptor */
+ /* Transiently set during Phase 2, normally set to NULL */
+ Relation rel;
/* Information saved by Phase 1 for Phase 2: */
List *subcmds[AT_NUM_PASSES]; /* Lists of AlterTableCmd */
/* Information saved by Phases 1/2 for Phase 3: */
@@ -355,7 +357,7 @@ static void ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
AlterTableUtilityContext *context);
static void ATRewriteCatalogs(List **wqueue, LOCKMODE lockmode,
AlterTableUtilityContext *context);
-static void ATExecCmd(List **wqueue, AlteredTableInfo *tab, Relation rel,
+static void ATExecCmd(List **wqueue, AlteredTableInfo *tab,
AlterTableCmd *cmd, LOCKMODE lockmode, int cur_pass,
AlterTableUtilityContext *context);
static AlterTableCmd *ATParseTransformCmd(List **wqueue, AlteredTableInfo *tab,
@@ -542,7 +544,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +554,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +996,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3168,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3367,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3986,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4384,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4407,7 +4429,6 @@ ATRewriteCatalogs(List **wqueue, LOCKMODE lockmode,
{
AlteredTableInfo *tab = (AlteredTableInfo *) lfirst(ltab);
List *subcmds = tab->subcmds[pass];
- Relation rel;
ListCell *lcmd;
if (subcmds == NIL)
@@ -4416,10 +4437,10 @@ ATRewriteCatalogs(List **wqueue, LOCKMODE lockmode,
/*
* Appropriate lock was obtained by phase 1, needn't get it again
*/
- rel = relation_open(tab->relid, NoLock);
+ tab->rel = relation_open(tab->relid, NoLock);
foreach(lcmd, subcmds)
- ATExecCmd(wqueue, tab, rel,
+ ATExecCmd(wqueue, tab,
castNode(AlterTableCmd, lfirst(lcmd)),
lockmode, pass, context);
@@ -4431,7 +4452,11 @@ ATRewriteCatalogs(List **wqueue, LOCKMODE lockmode,
if (pass == AT_PASS_ALTER_TYPE)
ATPostAlterTypeCleanup(wqueue, tab, lockmode);
- relation_close(rel, NoLock);
+ if (tab->rel)
+ {
+ relation_close(tab->rel, NoLock);
+ tab->rel = NULL;
+ }
}
}
@@ -4457,11 +4482,12 @@ ATRewriteCatalogs(List **wqueue, LOCKMODE lockmode,
* ATExecCmd: dispatch a subcommand to appropriate execution routine
*/
static void
-ATExecCmd(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ATExecCmd(List **wqueue, AlteredTableInfo *tab,
AlterTableCmd *cmd, LOCKMODE lockmode, int cur_pass,
AlterTableUtilityContext *context)
{
ObjectAddress address = InvalidObjectAddress;
+ Relation rel = tab->rel;
switch (cmd->subtype)
{
@@ -4748,7 +4774,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab, Relation rel,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -5564,6 +5595,7 @@ ATGetQueueEntry(List **wqueue, Relation rel)
*/
tab = (AlteredTableInfo *) palloc0(sizeof(AlteredTableInfo));
tab->relid = relid;
+ tab->rel = NULL; /* set later */
tab->relkind = rel->rd_rel->relkind;
tab->oldDesc = CreateTupleDescCopyConstr(RelationGetDescr(rel));
tab->newrelpersistence = RELPERSISTENCE_PERMANENT;
@@ -6179,7 +6211,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6412,7 +6444,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6566,7 +6599,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6666,17 +6699,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7275,7 +7313,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7858,7 +7896,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8322,7 +8361,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8937,7 +8977,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9071,7 +9111,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10852,7 +10892,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11236,7 +11277,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14131,7 +14173,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14142,12 +14184,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14161,7 +14261,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14177,7 +14277,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16285,7 +16387,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16345,7 +16447,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16933,105 +17035,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * in the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17104,22 +17331,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17315,7 +17670,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17469,7 +17824,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 092ac1646d..e461187b78 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1055,7 +1055,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1077,7 +1077,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 86594bd056..cc57dbf5a9 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -568,6 +568,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -578,7 +579,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1047,9 +1048,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1629,9 +1642,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1697,9 +1711,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 530aac68a7..c5d436dcfa 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4658,6 +4658,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index 0cf90ef33c..53287fe9a5 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2923,6 +2923,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 52c01eb86b..e56a853d87 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2134,10 +2134,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 357ab93fb6..a8e034df10 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -651,7 +651,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2053,12 +2053,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2066,8 +2067,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2082,6 +2096,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15140,6 +15155,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15678,6 +15694,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index ac0c495972..efb4a5e142 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..301d1c6844 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index f398027fa6..6c3caaa1ba 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1275,6 +1275,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index acf8a44f30..348b4f7dd2 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 07d640021c..9d2f1b562d 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2113,7 +2113,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2131,17 +2136,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3193,9 +3200,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3204,14 +3220,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3261,11 +3277,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index e1aeea2560..86d9839b1b 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1851,6 +1852,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 71dcdf2889..2d523568c1 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index b6b403e293..df40a3b4a4 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--HcAYCG3uE/tztfnV--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 28 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 7 +-
src/backend/commands/tablecmds.c | 568 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/cache/partcache.c | 15 +-
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 301 ++++++++++
.../detach-partition-concurrently-2.out | 66 ++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 90 +++
.../detach-partition-concurrently-2.spec | 41 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 79 +++
.../specs/partition-removal-1.spec | 45 ++
32 files changed, 1347 insertions(+), 173 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 3a2266526c..759dd8da5b 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4482,6 +4482,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index c25ef5abd6..09673dcf1c 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and its partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 21f2240ade..73e26fa400 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2546,10 +2546,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3818,7 +3820,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index cffbc0ac38..0dc53b25b8 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1831,7 +1831,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2425,7 +2425,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..4327a0b150 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,7 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,6 +42,9 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, InvalidOid is
+ * returned and no error is thrown.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
@@ -51,12 +54,15 @@ get_partition_parent(Oid relid)
{
Relation catalogRelation;
Oid result;
+ bool detached;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid, &detached);
- if (!OidIsValid(result))
+ if (detached)
+ result = InvalidOid;
+ else if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
table_close(catalogRelation, AccessShareLock);
@@ -70,13 +76,15 @@ get_partition_parent(Oid relid)
* given relation
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detached)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detached = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,7 +101,14 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
- result = form->inhparent;
+ /* A partition being detached has no parent */
+ if (form->inhdetached)
+ {
+ *detached = true;
+ result = InvalidOid;
+ }
+ else
+ result = form->inhparent;
}
systable_endscan(scan);
@@ -134,9 +149,10 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detached;
/* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
+ parentOid = get_partition_parent_worker(inhRel, relid, &detached);
if (parentOid == InvalidOid)
return;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index f3783961b7..3aec885288 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +209,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -428,6 +439,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -447,10 +459,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -477,6 +500,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 992f4813b4..ee4af9ea32 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -401,7 +401,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1114,7 +1114,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1171,7 +1171,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
@@ -3849,6 +3849,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index a8528a3423..d7b9c63e5f 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -542,7 +542,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -551,8 +552,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -987,7 +994,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3158,7 +3166,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3357,7 +3365,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3976,7 +3984,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4367,6 +4382,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4752,7 +4772,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6184,7 +6209,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6417,7 +6442,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6571,7 +6597,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6671,17 +6697,22 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
if (rel->rd_rel->relispartition)
{
Oid parentId = get_partition_parent(RelationGetRelid(rel));
- Relation parent = table_open(parentId, AccessShareLock);
- TupleDesc tupDesc = RelationGetDescr(parent);
+ Relation parent;
+ TupleDesc tupDesc;
AttrNumber parent_attnum;
- parent_attnum = get_attnum(parentId, colName);
- if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
- ereport(ERROR,
- (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
- errmsg("column \"%s\" is marked NOT NULL in parent table",
- colName)));
- table_close(parent, AccessShareLock);
+ if (parentId != InvalidOid)
+ {
+ parent = table_open(parentId, AccessShareLock);
+ tupDesc = RelationGetDescr(parent);
+ parent_attnum = get_attnum(parentId, colName);
+ if (TupleDescAttr(tupDesc, parent_attnum - 1)->attnotnull)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_TABLE_DEFINITION),
+ errmsg("column \"%s\" is marked NOT NULL in parent table",
+ colName)));
+ table_close(parent, AccessShareLock);
+ }
}
/*
@@ -7280,7 +7311,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -7863,7 +7894,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8327,7 +8359,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8942,7 +8975,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9076,7 +9109,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10860,7 +10893,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11244,7 +11278,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14139,7 +14174,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14150,12 +14185,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14169,7 +14262,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14185,7 +14278,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16293,7 +16388,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16353,7 +16448,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16942,105 +17037,230 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint, if there isn't a matching
+ * one already. The reason for this is that the planner may have made
+ * optimizations that depend on the constraint. XXX Isn't it sufficient to
+ * invalidate the partition's relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all processes to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint, if needed. XXX hopefully we can just remove this */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17113,22 +17333,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17324,7 +17672,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17478,7 +17826,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 12229364f1..d422d332d5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1129,7 +1129,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1151,7 +1151,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 941731a0a9..87336d81e0 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -567,6 +567,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -577,7 +578,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1046,9 +1047,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1628,9 +1641,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1696,9 +1710,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index ba3ccc712c..80685bdff9 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4678,6 +4678,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index a2ef853dc2..1e73308280 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2941,6 +2941,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index da322b453e..da5a8c5f5a 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2135,10 +2135,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 31c95443a5..4ae8caf79f 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -644,7 +644,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2094,12 +2094,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2107,8 +2108,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2123,6 +2137,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15276,6 +15291,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15814,6 +15830,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index b9aeb77bc2..336cc1db9b 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -4017,7 +4017,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4217,7 +4217,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 53a511f1da..38ff1e0923 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1274,6 +1274,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..74d85b89fb 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,19 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
+ /*
+ * Obtain parent relid; if it's invalid, then the partition is being
+ * detached
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel));
+ if (parentrelid == InvalidOid)
+ {
+ rel->rd_partcheckvalid = true;
+ return NIL;
+ }
+
/* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index caf97563f4..883ebdc474 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2117,7 +2117,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2135,17 +2140,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3197,9 +3204,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3208,14 +3224,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3265,11 +3281,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index a0f2ec4e0b..ba2a150128 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index dc2bb40926..1a2f87f5f9 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -860,6 +860,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1870,6 +1871,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 8c554e1f69..c79b76edca 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index 579be352c5..21487b6d44 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -133,6 +133,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..8d15beac60
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,301 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c s3debugtest
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2); <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3ins2: <... completed>
+step s3debugtest: SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2;
+tableoid
+
+d_listp1
+a
+
+2
+2
+2
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2detach s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2detach s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2detach s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2detach s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2detach s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2detach s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index f2e752c445..3c8fa4e07d 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..1e9bea8198
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,90 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+# XXX remove this step
+step "s3debugtest" { SELECT tableoid::regclass FROM d_listp; SELECT * from d_listp2; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c" "s3debugtest"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2detach" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2detach" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2detach" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2detach" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..00b93ecf65
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,79 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..e8e99f15bf
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,45 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway; however we expect
+# its rows to still appear in the result.
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2); }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3unlock" "s1c"
+
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
--
2.20.1
--/04w6evG8XlLl3ft--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-08-03 23:21 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-08-03 23:21 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 491 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 2 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 182 +++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 71 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 958 insertions(+), 153 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 26fda20d19..0f8abdc2d0 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4471,6 +4471,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index b2eb7097a9..2e67468e46 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index f2ca686397..7cc5cb4d51 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2552,10 +2552,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3787,7 +3789,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 1be27eec52..ad545a7866 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1655,7 +1655,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2241,7 +2241,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index 2baca12c5f..d032fe4af3 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1092,7 +1092,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1149,7 +1149,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3550,6 +3550,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 07f4f562ee..682aa6a015 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -541,7 +541,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +551,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +990,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3099,7 +3106,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3298,7 +3305,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3916,7 +3923,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4288,6 +4302,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4663,7 +4682,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -6081,7 +6105,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6314,7 +6338,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6468,7 +6493,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7680,7 +7705,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8144,7 +8170,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8759,7 +8786,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -8893,7 +8920,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10674,7 +10701,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11058,7 +11086,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -13953,7 +13982,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -13964,12 +13993,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -13983,7 +14070,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -13999,7 +14086,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16107,7 +16196,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16163,7 +16252,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16751,105 +16840,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -16922,17 +17132,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17133,7 +17424,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17287,7 +17578,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49056..ef4fd7bf90 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -507,6 +507,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -517,7 +518,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -997,7 +998,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1559,7 +1560,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1625,9 +1626,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409de66..f48350bc2f 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4670,6 +4670,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e3f33c40be..01ff684f90 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2932,6 +2932,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 25545029d7..75fe783b41 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2129,7 +2129,7 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
/* Create the PartitionDirectory infrastructure if we didn't already */
if (root->glob->partition_directory == NULL)
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index dbb47d4982..da0b233695 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15110,6 +15125,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7553d55987..966d4bbe7f 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2810,7 +2810,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3984,7 +3984,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4184,7 +4184,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..3c21f42ea9 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock,
+ include_detached);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9b0c376c8c..57ce760edc 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1240,6 +1240,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 57266f4fc3..329b650e12 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 151bcdb7ef..d291b4d7df 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1845,6 +1846,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..04a0446ccb
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,182 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec: EXECUTE f(1); DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1exec2: EXECUTE f(2); DEALLOCATE f;
+step s2d: <... completed>
+error in steps s1exec2 s2d: ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 218c87b24b..a9c1abc244 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..ca92530824
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,71 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1exec" { EXECUTE f(1); DEALLOCATE f; }
+step "s1exec2" { EXECUTE f(2); DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
--
2.20.1
--ikeVEW9yuYc//A+q--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2020-09-10 20:50 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2020-09-10 20:50 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 25 +-
src/backend/catalog/heap.c | 13 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/pg_inherits.c | 51 +-
src/backend/commands/indexcmds.c | 5 +-
src/backend/commands/tablecmds.c | 510 ++++++++++++++----
src/backend/commands/trigger.c | 7 +-
src/backend/executor/execPartition.c | 16 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 11 +-
src/backend/parser/gram.y | 22 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/bin/psql/describe.c | 40 +-
src/include/catalog/pg_inherits.h | 7 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
.../detach-partition-concurrently-1.out | 233 ++++++++
.../detach-partition-concurrently-2.out | 66 +++
src/test/isolation/isolation_schedule | 2 +
.../detach-partition-concurrently-1.spec | 76 +++
.../detach-partition-concurrently-2.spec | 41 ++
26 files changed, 1035 insertions(+), 160 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index 508bea3bc6..abef70ad46 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4481,6 +4481,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetached</structfield> <type>bool</type>
+ </para>
+ <para>
+ Set to true for a partition that is in the process of being detached;
+ false otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index e9c6a8a6c1..c35e83f229 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -938,7 +940,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -947,6 +950,24 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
</para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction,
+ <literal>SHARE UPDATE EXCLUSIVE</literal> is taken in both parent table and
+ partition, and the partition is marked detached; at that point, the transaction
+ is committed and all transactions using the partitioned table are waited for.
+ Once all those transactions are gone, the second stage acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, complete actions of a
+ previous <literal>DETACH CONCURRENTLY</literal> invocation that
+ was cancelled or crashed.
+ </para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index 67144aa3c9..e4a4064511 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -2544,10 +2544,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3774,7 +3776,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 117e3fdef7..ffd2c2770f 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1669,7 +1669,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2255,7 +2255,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 17f37eb39f..f143db0ad4 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -52,7 +52,8 @@ typedef struct SeenRelsEntry
* against possible DROPs of child relations.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,7 +92,13 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /* skip detached at caller's request */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached &&
+ !include_detached)
+ continue;
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
+
if (numoids >= maxoids)
{
maxoids *= 2;
@@ -160,6 +167,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -200,7 +210,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -429,6 +440,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -448,10 +460,21 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * 'detached' is the expected state of the inhdetached flag. If the catalog
+ * row does not match that state, an error is raised. When used on
+ * partitions, the partition name must be passed, for possible error messages.
+ *
+ * If allow_detached is passed false, then an error is raised if the child is
+ * already marked detached. A name must be supplied in that case, for the
+ * error message. This should only be used with table partitions.
+ * XXX the name bit is pretty weird and problematic for non-partition cases;
+ * what if the flag check fails?
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool detached,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -478,6 +501,28 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool inhdetached;
+
+ inhdetached = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetached;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (inhdetached && !detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!inhdetached && detached)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname),
+ errdetail("There's no partial concurrent detach in progress.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index f1b5f87e6a..424eb6a19b 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -1105,7 +1105,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1162,7 +1162,7 @@ DefineIndex(Oid relationId,
*/
if (!stmt->relation || stmt->relation->inh)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int nparts = partdesc->nparts;
Oid *part_oids = palloc(sizeof(Oid) * nparts);
bool invalidate_parent = false;
@@ -3751,6 +3751,7 @@ IndexSetParentIndex(Relation partitionIdx, Oid parentOid)
values[Anum_pg_inherits_inhparent - 1] =
ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(1);
+ values[Anum_pg_inherits_inhdetached - 1] = BoolGetDatum(false);
memset(isnull, false, sizeof(isnull));
tuple = heap_form_tuple(RelationGetDescr(pg_inherits),
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 3e57c7f9e1..46a84fddfb 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -157,6 +157,8 @@ typedef struct AlteredTableInfo
Oid relid; /* Relation to work on */
char relkind; /* Its relkind */
TupleDesc oldDesc; /* Pre-modification tuple descriptor */
+ /* Transiently set during Phase 2, normally set to NULL */
+ Relation rel;
/* Information saved by Phase 1 for Phase 2: */
List *subcmds[AT_NUM_PASSES]; /* Lists of AlterTableCmd */
/* Information saved by Phases 1/2 for Phase 3: */
@@ -354,7 +356,7 @@ static void ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
AlterTableUtilityContext *context);
static void ATRewriteCatalogs(List **wqueue, LOCKMODE lockmode,
AlterTableUtilityContext *context);
-static void ATExecCmd(List **wqueue, AlteredTableInfo *tab, Relation rel,
+static void ATExecCmd(List **wqueue, AlteredTableInfo *tab,
AlterTableCmd *cmd, LOCKMODE lockmode, int cur_pass,
AlterTableUtilityContext *context);
static AlterTableCmd *ATParseTransformCmd(List **wqueue, AlteredTableInfo *tab,
@@ -541,7 +543,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd);
static void AttachPartitionEnsureIndexes(Relation rel, Relation attachrel);
@@ -550,7 +553,12 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -984,7 +992,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -3152,7 +3161,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3351,7 +3360,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -3970,7 +3979,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4349,6 +4365,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4389,7 +4410,6 @@ ATRewriteCatalogs(List **wqueue, LOCKMODE lockmode,
{
AlteredTableInfo *tab = (AlteredTableInfo *) lfirst(ltab);
List *subcmds = tab->subcmds[pass];
- Relation rel;
ListCell *lcmd;
if (subcmds == NIL)
@@ -4398,10 +4418,10 @@ ATRewriteCatalogs(List **wqueue, LOCKMODE lockmode,
/*
* Appropriate lock was obtained by phase 1, needn't get it again
*/
- rel = relation_open(tab->relid, NoLock);
+ tab->rel = relation_open(tab->relid, NoLock);
foreach(lcmd, subcmds)
- ATExecCmd(wqueue, tab, rel,
+ ATExecCmd(wqueue, tab,
castNode(AlterTableCmd, lfirst(lcmd)),
lockmode, pass, context);
@@ -4413,7 +4433,11 @@ ATRewriteCatalogs(List **wqueue, LOCKMODE lockmode,
if (pass == AT_PASS_ALTER_TYPE)
ATPostAlterTypeCleanup(wqueue, tab, lockmode);
- relation_close(rel, NoLock);
+ if (tab->rel)
+ {
+ relation_close(tab->rel, NoLock);
+ tab->rel = NULL;
+ }
}
}
@@ -4439,11 +4463,12 @@ ATRewriteCatalogs(List **wqueue, LOCKMODE lockmode,
* ATExecCmd: dispatch a subcommand to appropriate execution routine
*/
static void
-ATExecCmd(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ATExecCmd(List **wqueue, AlteredTableInfo *tab,
AlterTableCmd *cmd, LOCKMODE lockmode, int cur_pass,
AlterTableUtilityContext *context)
{
ObjectAddress address = InvalidObjectAddress;
+ Relation rel = tab->rel;
switch (cmd->subtype)
{
@@ -4729,7 +4754,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab, Relation rel,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
@@ -5540,6 +5570,7 @@ ATGetQueueEntry(List **wqueue, Relation rel)
*/
tab = (AlteredTableInfo *) palloc0(sizeof(AlteredTableInfo));
tab->relid = relid;
+ tab->rel = NULL; /* set later */
tab->relkind = rel->rd_rel->relkind;
tab->oldDesc = CreateTupleDescCopyConstr(RelationGetDescr(rel));
tab->newrelpersistence = RELPERSISTENCE_PERMANENT;
@@ -6159,7 +6190,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6392,7 +6423,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6546,7 +6578,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -7787,7 +7819,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8251,7 +8284,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -8866,7 +8900,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9000,7 +9034,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -10781,7 +10815,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11165,7 +11200,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14060,7 +14096,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14071,12 +14107,70 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * When a partition is detached from its parent concurrently, we don't
+ * remove the pg_inherits row until a second transaction; as a preparatory
+ * step, this function marks the entry as 'detached', so that other
+ * transactions know to ignore the partition.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetached = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14090,7 +14184,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14106,7 +14200,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16214,7 +16310,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16270,7 +16366,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd)
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -16858,105 +16954,226 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition catalog
+ * rows to make it visible to everyone that the partition is detached,
+ * lock the partition against writes, and commit the transaction; anyone
+ * who requests the partition descriptor from that point onwards has to
+ * ignore such a partition. In a second transaction, we wait until all
+ * transactions that could have seen the partition as attached are gone,
+ * then we remove the rest of partition metadata (pg_inherits and
+ * pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with
+ * ShareUpdateExclusiveLock in the first stage.
+ */
+ partRel = table_openrv(name, concurrent ?
+ ShareUpdateExclusiveLock : AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhisdetached flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps lock
+ * on the referencing tables, which prevent concurrent transactions from
+ * adding rows that we wouldn't see. For this to work in concurrent mode,
+ * it is critical that the partition appears as no longer attached for the
+ * RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to the
+ * partition that matches the partition constraint. The reason for this is
+ * that the planner may have made optimizations that depend on the
+ * constraint. XXX Isn't it sufficient to invalidate the partition's
+ * relcache entry?
+ *
+ * Then we close our existing transaction, and in a new one wait for
+ * all process to catch up on the catalog updates we've done so far; at
+ * that point we can complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ Constraint *n;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(make_ands_explicit(RelationGetPartitionQual(partRel)));
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17029,17 +17246,98 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
@@ -17240,7 +17538,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17394,7 +17692,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..ce65abd57e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1054,7 +1054,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1076,7 +1076,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
@@ -1537,7 +1538,7 @@ EnableDisableTrigger(Relation rel, const char *tgname,
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
(TRIGGER_FOR_ROW(oldtrig->tgtype)))
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index bd2ea25804..e4779c5433 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -564,6 +564,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation rootrel = rootResultRelInfo->ri_RelationDesc,
partrel;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -574,7 +575,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1054,7 +1055,7 @@ ExecInitPartitionDispatchInfo(EState *estate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1636,7 +1637,7 @@ ExecCreatePartitionPruneState(PlanState *planstate,
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1702,9 +1703,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 0409a40b82..044488f14c 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4669,6 +4669,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index e2d1b987bf..8fb1d082e5 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2931,6 +2931,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index f9d0d67aa7..0fe053be96 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2123,10 +2123,17 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ * In snapshot-isolation transactions, we always include detached
+ * partitions, because otherwise they become invisible too soon.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext,
+ IsolationUsesXactSnapshot());
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index c5154b818c..3bfbfdd8a5 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -649,7 +649,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2057,12 +2057,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2070,8 +2071,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2086,6 +2100,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15114,6 +15129,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 58f9b46289..0e872b9283 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2808,7 +2808,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
ParseState *pstate = make_parsestate(NULL);
int with = -1;
@@ -3982,7 +3982,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4182,7 +4182,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index 0f124a5254..301d1c6844 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -38,6 +38,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -47,7 +48,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -62,13 +63,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -89,7 +91,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -111,7 +113,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -239,6 +242,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -281,7 +285,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -296,6 +300,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_mcxt = mcxt;
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -328,7 +333,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 9713a7ac41..81d3c15de8 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1264,6 +1264,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index 0861d74a6f..86c23ea66a 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2109,7 +2109,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetached" :
+ "false as inhdetached");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2127,17 +2132,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACHED" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3179,9 +3186,18 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, inhdetached,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3190,14 +3206,14 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind, false AS inhdetached, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3247,11 +3263,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACHED)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 0432c3ab88..fcdea9f26f 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid;
Oid inhparent;
int32 inhseqno;
+ bool inhdetached;
} FormData_pg_inherits;
/* ----------------
@@ -44,7 +45,8 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
typedef FormData_pg_inherits *Form_pg_inherits;
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -52,6 +54,7 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index e83329fd6d..939ccb4d4a 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -859,6 +859,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1846,6 +1847,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index 08f22ce211..27414b0912 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -163,6 +163,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index 70df764981..b1ee9fd5e3 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..ce493e3b93
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,233 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s1s s1c s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2d: <... completed>
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2d s1ins s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1ins s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1s s2d s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec: EXECUTE f(1);
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1brr s1prep s1s s2d s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2d: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2d s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2d: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2d s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2d: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1b s1s s2d s3drop1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop1: DROP TABLE d_listp; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop1: <... completed>
+error in steps s1c s2d s3drop1: ERROR: partitioned table "d_listp" was removed concurrently
+
+starting permutation: s1b s1s s2d s3drop2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3drop2: DROP TABLE d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3drop2: <... completed>
+error in steps s1c s2d s3drop2: ERROR: partition "d_listp2" was removed concurrently
+
+starting permutation: s1b s1s s2d s3detach s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3detach: ALTER TABLE d_listp DETACH PARTITION d_listp2; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3detach: <... completed>
+error in steps s1c s2d s3detach: ERROR: cannot detach partition "d_listp2"
+
+starting permutation: s1b s1s s2d s3rename s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2d: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3rename: ALTER TABLE d_listp2 RENAME TO d_listp_foobar; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3rename: <... completed>
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index aa386ab1a2..c381fcd967 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,8 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..f035db8c5d
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,76 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1exec" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3drop1" { DROP TABLE d_listp; }
+step "s3drop2" { DROP TABLE d_listp2; }
+step "s3detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2; }
+step "s3rename" { ALTER TABLE d_listp2 RENAME TO d_listp_foobar; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s1s" "s1c" "s2drop" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2d" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# Session s1 no longer sees the partition ...
+permutation "s1b" "s1s" "s2d" "s1ins" "s1c"
+# ... regardless of the isolation level.
+permutation "s1brr" "s1s" "s2d" "s1ins" "s1c"
+# However, DETACH CONCURRENTLY is not isolation-level-correct
+permutation "s1brr" "s1s" "s2d" "s1s" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1s" "s2d" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2d" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2d" "s1s" "s1exec2" "s1c" "s1dealloc"
+
+
+# Note: the following tests are not very interesting in isolationtester because
+# the way this tool handles multiple sessions: it'll only run the drop/detach/
+# rename commands after s2 is blocked waiting. It would be more useful to
+# run the DDL when s2 commits its first transaction instead. Maybe these should
+# become TAP tests someday.
+
+# What happens if the partition or the parent table is dropped while waiting?
+permutation "s1b" "s1s" "s2d" "s3drop1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3drop2" "s1c"
+# Also try a non-concurrent detach while the other one is waiting
+permutation "s1b" "s1s" "s2d" "s3detach" "s1c"
+# and some other DDL
+permutation "s1b" "s1s" "s2d" "s3rename" "s1c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
base-commit: 58c6feccfae1321ea4ff16c535c6f3fb90a07d69
--
2.20.1
--Kj7319i9nmIyA2yE--
^ permalink raw reply [nested|flat] 57+ messages in thread
* [PATCH v9 2/2] ALTER TABLE ... DETACH CONCURRENTLY
@ 2021-03-25 15:44 Alvaro Herrera <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Alvaro Herrera @ 2021-03-25 15:44 UTC (permalink / raw)
---
doc/src/sgml/catalogs.sgml | 10 +
doc/src/sgml/ref/alter_table.sgml | 28 +-
src/backend/catalog/heap.c | 20 +-
src/backend/catalog/index.c | 4 +-
src/backend/catalog/partition.c | 38 +-
src/backend/catalog/pg_inherits.c | 114 +++-
src/backend/commands/indexcmds.c | 6 +-
src/backend/commands/tablecmds.c | 569 ++++++++++++++----
src/backend/commands/trigger.c | 5 +-
src/backend/executor/execPartition.c | 29 +-
src/backend/nodes/copyfuncs.c | 1 +
src/backend/nodes/equalfuncs.c | 1 +
src/backend/optimizer/util/plancat.c | 8 +-
src/backend/parser/gram.y | 23 +-
src/backend/partitioning/partbounds.c | 6 +-
src/backend/partitioning/partdesc.c | 21 +-
src/backend/tcop/utility.c | 19 +
src/backend/utils/adt/ri_triggers.c | 6 +-
src/backend/utils/cache/partcache.c | 12 +-
src/bin/psql/describe.c | 42 +-
src/include/catalog/partition.h | 2 +-
src/include/catalog/pg_inherits.h | 8 +-
src/include/nodes/parsenodes.h | 2 +
src/include/parser/kwlist.h | 1 +
src/include/partitioning/partdesc.h | 5 +-
src/include/utils/snapmgr.h | 1 +
.../detach-partition-concurrently-1.out | 234 +++++++
.../detach-partition-concurrently-2.out | 66 ++
.../detach-partition-concurrently-3.out | 282 +++++++++
.../detach-partition-concurrently-4.out | 69 +++
src/test/isolation/isolation_schedule | 4 +
.../detach-partition-concurrently-1.spec | 69 +++
.../detach-partition-concurrently-2.spec | 41 ++
.../detach-partition-concurrently-3.spec | 66 ++
.../detach-partition-concurrently-4.spec | 36 ++
src/test/modules/delay_execution/Makefile | 3 +-
.../expected/partition-removal-1.out | 175 ++++++
.../specs/partition-removal-1.spec | 58 ++
src/test/regress/expected/alter_table.out | 29 +
src/test/regress/sql/alter_table.sql | 20 +
40 files changed, 1959 insertions(+), 174 deletions(-)
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-1.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-2.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-3.out
create mode 100644 src/test/isolation/expected/detach-partition-concurrently-4.out
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-1.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-2.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-3.spec
create mode 100644 src/test/isolation/specs/detach-partition-concurrently-4.spec
create mode 100644 src/test/modules/delay_execution/expected/partition-removal-1.out
create mode 100644 src/test/modules/delay_execution/specs/partition-removal-1.spec
diff --git a/doc/src/sgml/catalogs.sgml b/doc/src/sgml/catalogs.sgml
index bae4d8cdd3..cd00d9e3bb 100644
--- a/doc/src/sgml/catalogs.sgml
+++ b/doc/src/sgml/catalogs.sgml
@@ -4497,6 +4497,16 @@ SCRAM-SHA-256$<replaceable><iteration count></replaceable>:<replaceable>&l
when using declarative partitioning.
</para></entry>
</row>
+
+ <row>
+ <entry role="catalog_table_entry"><para role="column_definition">
+ <structfield>inhdetachpending</structfield> <type>bool</type>
+ </para>
+ <para>
+ <literal>true</literal> for a partition that is in the process of
+ being detached; <literal>false</literal> otherwise.
+ </para></entry>
+ </row>
</tbody>
</tgroup>
</table>
diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml
index 0bd0c1a503..6db505af45 100644
--- a/doc/src/sgml/ref/alter_table.sgml
+++ b/doc/src/sgml/ref/alter_table.sgml
@@ -36,7 +36,9 @@ ALTER TABLE ALL IN TABLESPACE <replaceable class="parameter">name</replaceable>
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
ATTACH PARTITION <replaceable class="parameter">partition_name</replaceable> { FOR VALUES <replaceable class="parameter">partition_bound_spec</replaceable> | DEFAULT }
ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
- DETACH PARTITION <replaceable class="parameter">partition_name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> [ CONCURRENTLY ]
+ALTER TABLE [ IF EXISTS ] <replaceable class="parameter">name</replaceable>
+ DETACH PARTITION <replaceable class="parameter">partition_name</replaceable> FINALIZE
<phrase>where <replaceable class="parameter">action</replaceable> is one of:</phrase>
@@ -954,7 +956,8 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
</varlistentry>
<varlistentry>
- <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable></term>
+ <term><literal>DETACH PARTITION</literal> <replaceable class="parameter">partition_name</replaceable> [ { CONCURRENTLY | FINALIZE } ]</term>
+
<listitem>
<para>
This form detaches the specified partition of the target table. The detached
@@ -962,6 +965,27 @@ WITH ( MODULUS <replaceable class="parameter">numeric_literal</replaceable>, REM
ties to the table from which it was detached. Any indexes that were
attached to the target table's indexes are detached. Any triggers that
were created as clones of those in the target table are removed.
+ <literal>SHARE<literal> lock is obtained on any tables that reference
+ this partitioned table in foreign key constraints.
+ </para>
+ <para>
+ If <literal>CONCURRENTLY</literal> is specified, this process runs in two
+ transactions in order to avoid blocking other sessions that might be accessing
+ the partitioned table. During the first transaction, a
+ <literal>SHARE UPDATE EXCLUSIVE</literal> lock is taken on both parent table and
+ partition, and the partition is marked as undergoing detach;
+ at that point, the transaction is committed and all other transactions using
+ the partitioned table are waited for.
+ Once all those transactions have completed, the second transaction acquires
+ <literal>ACCESS EXCLUSIVE</literal> on the partition, and the detach process
+ completes.
+ <literal>CONCURRENTLY</literal> is not allowed if the
+ partitioned table contains a default partition.
+ </para>
+ <para>
+ If <literal>FINALIZE</literal> is specified, a previous
+ <literal>DETACH CONCURRENTLY</literal> invocation that was cancelled or
+ interrupted is completed.
</para>
</listitem>
</varlistentry>
diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c
index d0ec44bb40..9f6303266f 100644
--- a/src/backend/catalog/heap.c
+++ b/src/backend/catalog/heap.c
@@ -1923,7 +1923,12 @@ heap_drop_with_catalog(Oid relid)
elog(ERROR, "cache lookup failed for relation %u", relid);
if (((Form_pg_class) GETSTRUCT(tuple))->relispartition)
{
- parentOid = get_partition_parent(relid);
+ /*
+ * We have to lock the parent if the partition is being detached,
+ * because it's possible that some query still has a partition
+ * descriptor that includes this partition.
+ */
+ parentOid = get_partition_parent(relid, true);
LockRelationOid(parentOid, AccessExclusiveLock);
/*
@@ -2559,10 +2564,12 @@ StoreConstraints(Relation rel, List *cooked_constraints, bool is_internal)
* Returns a list of CookedConstraint nodes that shows the cooked form of
* the default and constraint expressions added to the relation.
*
- * NB: caller should have opened rel with AccessExclusiveLock, and should
- * hold that lock till end of transaction. Also, we assume the caller has
- * done a CommandCounterIncrement if necessary to make the relation's catalog
- * tuples visible.
+ * NB: caller should have opened rel with some self-conflicting lock mode,
+ * and should hold that lock till end of transaction; for normal cases that'll
+ * be AccessExclusiveLock, but if caller knows that the constraint is already
+ * enforced by some other means, it can be ShareUpdateExclusiveLock. Also, we
+ * assume the caller has done a CommandCounterIncrement if necessary to make
+ * the relation's catalog tuples visible.
*/
List *
AddRelationNewConstraints(Relation rel,
@@ -3831,7 +3838,8 @@ StorePartitionBound(Relation rel, Relation parent, PartitionBoundSpec *bound)
* relcache entry for that partition every time a partition is added or
* removed.
*/
- defaultPartOid = get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ defaultPartOid =
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent, false));
if (OidIsValid(defaultPartOid))
CacheInvalidateRelcacheByRelid(defaultPartOid);
diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c
index 397d70d226..af30840c04 100644
--- a/src/backend/catalog/index.c
+++ b/src/backend/catalog/index.c
@@ -1837,7 +1837,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName)
List *ancestors = get_partition_ancestors(oldIndexId);
Oid parentIndexRelid = linitial_oid(ancestors);
- DeleteInheritsTuple(oldIndexId, parentIndexRelid);
+ DeleteInheritsTuple(oldIndexId, parentIndexRelid, false, NULL);
StoreSingleInheritance(newIndexId, parentIndexRelid, 1);
list_free(ancestors);
@@ -2487,7 +2487,7 @@ index_drop(Oid indexId, bool concurrent, bool concurrent_lock_mode)
/*
* fix INHERITS relation
*/
- DeleteInheritsTuple(indexId, InvalidOid);
+ DeleteInheritsTuple(indexId, InvalidOid, false, NULL);
/*
* We are presently too lazy to attempt to compute the new correct value
diff --git a/src/backend/catalog/partition.c b/src/backend/catalog/partition.c
index af7754d6ab..790f4ccb92 100644
--- a/src/backend/catalog/partition.c
+++ b/src/backend/catalog/partition.c
@@ -32,7 +32,8 @@
#include "utils/rel.h"
#include "utils/syscache.h"
-static Oid get_partition_parent_worker(Relation inhRel, Oid relid);
+static Oid get_partition_parent_worker(Relation inhRel, Oid relid,
+ bool *detach_pending);
static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
List **ancestors);
@@ -42,23 +43,32 @@ static void get_partition_ancestors_worker(Relation inhRel, Oid relid,
*
* Returns inheritance parent of a partition by scanning pg_inherits
*
+ * If the partition is in the process of being detached, an error is thrown,
+ * unless even_if_detached is passed as true.
+ *
* Note: Because this function assumes that the relation whose OID is passed
* as an argument will have precisely one parent, it should only be called
* when it is known that the relation is a partition.
*/
Oid
-get_partition_parent(Oid relid)
+get_partition_parent(Oid relid, bool even_if_detached)
{
Relation catalogRelation;
Oid result;
+ bool detach_pending;
catalogRelation = table_open(InheritsRelationId, AccessShareLock);
- result = get_partition_parent_worker(catalogRelation, relid);
+ result = get_partition_parent_worker(catalogRelation, relid,
+ &detach_pending);
if (!OidIsValid(result))
elog(ERROR, "could not find tuple for parent of relation %u", relid);
+ if (detach_pending && !even_if_detached)
+ elog(ERROR, "relation %u has no parent because it's being detached",
+ relid);
+
table_close(catalogRelation, AccessShareLock);
return result;
@@ -68,15 +78,20 @@ get_partition_parent(Oid relid)
* get_partition_parent_worker
* Scan the pg_inherits relation to return the OID of the parent of the
* given relation
+ *
+ * If the partition is being detached, *detach_pending is set true (but the
+ * original parent is still returned.)
*/
static Oid
-get_partition_parent_worker(Relation inhRel, Oid relid)
+get_partition_parent_worker(Relation inhRel, Oid relid, bool *detach_pending)
{
SysScanDesc scan;
ScanKeyData key[2];
Oid result = InvalidOid;
HeapTuple tuple;
+ *detach_pending = false;
+
ScanKeyInit(&key[0],
Anum_pg_inherits_inhrelid,
BTEqualStrategyNumber, F_OIDEQ,
@@ -93,6 +108,9 @@ get_partition_parent_worker(Relation inhRel, Oid relid)
{
Form_pg_inherits form = (Form_pg_inherits) GETSTRUCT(tuple);
+ /* Let caller know of partition being detached */
+ if (form->inhdetachpending)
+ *detach_pending = true;
result = form->inhparent;
}
@@ -134,10 +152,14 @@ static void
get_partition_ancestors_worker(Relation inhRel, Oid relid, List **ancestors)
{
Oid parentOid;
+ bool detach_pending;
- /* Recursion ends at the topmost level, ie., when there's no parent */
- parentOid = get_partition_parent_worker(inhRel, relid);
- if (parentOid == InvalidOid)
+ /*
+ * Recursion ends at the topmost level, ie., when there's no parent; also
+ * when the partition is being detached.
+ */
+ parentOid = get_partition_parent_worker(inhRel, relid, &detach_pending);
+ if (parentOid == InvalidOid || detach_pending)
return;
*ancestors = lappend_oid(*ancestors, parentOid);
@@ -170,7 +192,7 @@ index_get_partition(Relation partition, Oid indexId)
ReleaseSysCache(tup);
if (!ispartition)
continue;
- if (get_partition_parent(partIdx) == indexId)
+ if (get_partition_parent(partIdx, false) == indexId)
{
list_free(idxlist);
return partIdx;
diff --git a/src/backend/catalog/pg_inherits.c b/src/backend/catalog/pg_inherits.c
index 5ab7902827..bedee069be 100644
--- a/src/backend/catalog/pg_inherits.c
+++ b/src/backend/catalog/pg_inherits.c
@@ -29,6 +29,7 @@
#include "utils/builtins.h"
#include "utils/fmgroids.h"
#include "utils/memutils.h"
+#include "utils/snapmgr.h"
#include "utils/syscache.h"
/*
@@ -50,9 +51,14 @@ typedef struct SeenRelsEntry
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * include_detached says to include all partitions, even if they're marked
+ * detached. Passing it as false means they might or might not be included,
+ * depending on the visibility of the pg_inherits row for the active snapshot.
*/
List *
-find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
+find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode)
{
List *list = NIL;
Relation relation;
@@ -91,6 +97,30 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
while ((inheritsTuple = systable_getnext(scan)) != NULL)
{
+ /*
+ * Cope with partitions concurrently being detached. When we see a
+ * partition marked "detach pending", we only include it in the set of
+ * visible partitions if caller requested all detached partitions, or
+ * if its pg_inherits tuple's xmin is still visible to the active
+ * snapshot.
+ *
+ * The reason for this check is that we want to avoid seeing the
+ * partition as alive in RI queries during REPEATABLE READ or
+ * SERIALIZABLE transactions.
+ */
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetachpending &&
+ !include_detached)
+ {
+ TransactionId xmin;
+ Snapshot snap;
+
+ xmin = HeapTupleHeaderGetXmin(inheritsTuple->t_data);
+ snap = GetActiveSnapshot();
+
+ if (!XidInMVCCSnapshot(xmin, snap))
+ continue;
+ }
+
inhrelid = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhrelid;
if (numoids >= maxoids)
{
@@ -160,6 +190,9 @@ find_inheritance_children(Oid parentrelId, LOCKMODE lockmode)
* given rel; caller should already have locked it). If lockmode is NoLock
* then no locks are acquired, but caller must beware of race conditions
* against possible DROPs of child relations.
+ *
+ * NB - No current callers of this routine are interested in children being
+ * concurrently detached, so there's no provision to include them.
*/
List *
find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
@@ -199,7 +232,8 @@ find_all_inheritors(Oid parentrelId, LOCKMODE lockmode, List **numparents)
ListCell *lc;
/* Get the direct children of this rel */
- currentchildren = find_inheritance_children(currentrel, lockmode);
+ currentchildren = find_inheritance_children(currentrel, false,
+ lockmode);
/*
* Add to the queue only those children not already seen. This avoids
@@ -430,6 +464,7 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
values[Anum_pg_inherits_inhrelid - 1] = ObjectIdGetDatum(relationId);
values[Anum_pg_inherits_inhparent - 1] = ObjectIdGetDatum(parentOid);
values[Anum_pg_inherits_inhseqno - 1] = Int32GetDatum(seqNumber);
+ values[Anum_pg_inherits_inhdetachpending - 1] = BoolGetDatum(false);
memset(nulls, 0, sizeof(nulls));
@@ -449,10 +484,17 @@ StoreSingleInheritance(Oid relationId, Oid parentOid, int32 seqNumber)
* as InvalidOid, in which case all tuples matching inhrelid are deleted;
* otherwise only delete tuples with the specified inhparent.
*
+ * expect_detach_pending is the expected state of the inhdetachpending flag.
+ * If the catalog row does not match that state, an error is raised.
+ *
+ * childname is the partition name, if a table; pass NULL for regular
+ * inheritance or when working with other relation kinds.
+ *
* Returns whether at least one row was deleted.
*/
bool
-DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
+DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool expect_detach_pending,
+ const char *childname)
{
bool found = false;
Relation catalogRelation;
@@ -479,6 +521,29 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
parent = ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent;
if (!OidIsValid(inhparent) || parent == inhparent)
{
+ bool detach_pending;
+
+ detach_pending =
+ ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetachpending;
+
+ /*
+ * Raise error depending on state. This should only happen for
+ * partitions, but we have no way to cross-check.
+ */
+ if (detach_pending && !expect_detach_pending)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partition \"%s\"",
+ childname ? childname : "unknown relation"),
+ errdetail("The partition is being detached concurrently or has an unfinished detach."),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation")));
+ if (!detach_pending && expect_detach_pending)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot complete detaching partition \"%s\"",
+ childname ? childname : "unknown relation"),
+ errdetail("There's no pending concurrent detach.")));
+
CatalogTupleDelete(catalogRelation, &inheritsTuple->t_self);
found = true;
}
@@ -490,3 +555,46 @@ DeleteInheritsTuple(Oid inhrelid, Oid inhparent)
return found;
}
+
+/*
+ * Return whether the pg_inherits tuple for a partition has the "detach
+ * pending" flag set.
+ */
+bool
+PartitionHasPendingDetach(Oid partoid)
+{
+ Relation catalogRelation;
+ ScanKeyData key;
+ SysScanDesc scan;
+ HeapTuple inheritsTuple;
+
+ /* We don't have a good way to verify it is in fact a partition */
+
+ /*
+ * Find the pg_inherits entry by inhrelid. (There should only be one.)
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(partoid));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ bool detached;
+
+ detached =
+ ((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhdetachpending;
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ return detached;
+ }
+
+ elog(ERROR, "relation %u is not a partition", partoid);
+ return false; /* keep compiler quiet */
+}
diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c
index e4e1bbb7e0..166374cc0c 100644
--- a/src/backend/commands/indexcmds.c
+++ b/src/backend/commands/indexcmds.c
@@ -410,7 +410,7 @@ CompareOpclassOptions(Datum *opts1, Datum *opts2, int natts)
* GetCurrentVirtualXIDs. If, during any iteration, a particular vxid
* doesn't show up in the output, we know we can forget about it.
*/
-static void
+void
WaitForOlderSnapshots(TransactionId limitXmin, bool progress)
{
int n_old_snapshots;
@@ -1123,7 +1123,7 @@ DefineIndex(Oid relationId,
*/
if (partitioned && stmt->relation && !stmt->relation->inh)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
if (pd->nparts != 0)
flags |= INDEX_CREATE_INVALID;
@@ -1180,7 +1180,7 @@ DefineIndex(Oid relationId,
*
* If we're called internally (no stmt->relation), recurse always.
*/
- partdesc = RelationGetPartitionDesc(rel);
+ partdesc = RelationGetPartitionDesc(rel, false);
if ((!stmt->relation || stmt->relation->inh) && partdesc->nparts > 0)
{
int nparts = partdesc->nparts;
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index bf7fd6e8ae..0966636fb2 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -544,7 +544,8 @@ static PartitionSpec *transformPartitionSpec(Relation rel, PartitionSpec *partsp
static void ComputePartitionAttrs(ParseState *pstate, Relation rel, List *partParams, AttrNumber *partattrs,
List **partexprs, Oid *partopclass, Oid *partcollation, char strategy);
static void CreateInheritance(Relation child_rel, Relation parent_rel);
-static void RemoveInheritance(Relation child_rel, Relation parent_rel);
+static void RemoveInheritance(Relation child_rel, Relation parent_rel,
+ bool allow_detached);
static ObjectAddress ATExecAttachPartition(List **wqueue, Relation rel,
PartitionCmd *cmd,
AlterTableUtilityContext *context);
@@ -553,8 +554,14 @@ static void QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
List *partConstraint,
bool validate_default);
static void CloneRowTriggersToPartition(Relation parent, Relation partition);
+static void DetachAddConstraintIfNeeded(List **wqueue, Relation partRel);
static void DropClonedTriggersFromPartition(Oid partitionId);
-static ObjectAddress ATExecDetachPartition(Relation rel, RangeVar *name);
+static ObjectAddress ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab,
+ Relation rel, RangeVar *name,
+ bool concurrent);
+static void DetachPartitionFinalize(Relation rel, Relation partRel,
+ bool concurrent, Oid defaultPartOid, bool wait);
+static ObjectAddress ATExecDetachPartitionFinalize(Relation rel, RangeVar *name);
static ObjectAddress ATExecAttachPartitionIdx(List **wqueue, Relation rel,
RangeVar *name);
static void validatePartitionedIndex(Relation partedIdx, Relation partedTbl);
@@ -1002,7 +1009,8 @@ DefineRelation(CreateStmt *stmt, char relkind, Oid ownerId,
* lock the partition so as to avoid a deadlock.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(parent));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(parent,
+ false));
if (OidIsValid(defaultPartOid))
defaultRel = table_open(defaultPartOid, AccessExclusiveLock);
@@ -1555,7 +1563,7 @@ RangeVarCallbackForDropRelation(const RangeVar *rel, Oid relOid, Oid oldRelOid,
*/
if (is_partition && relOid != oldRelOid)
{
- state->partParentOid = get_partition_parent(relOid);
+ state->partParentOid = get_partition_parent(relOid, true);
if (OidIsValid(state->partParentOid))
LockRelationOid(state->partParentOid, AccessExclusiveLock);
}
@@ -3315,7 +3323,7 @@ renameatt_internal(Oid myrelid,
* expected_parents will only be 0 if we are not already recursing.
*/
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited column \"%s\" must be renamed in child tables too",
@@ -3514,7 +3522,7 @@ rename_constraint_internal(Oid myrelid,
else
{
if (expected_parents == 0 &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("inherited constraint \"%s\" must be renamed in child tables too",
@@ -4134,7 +4142,14 @@ AlterTableGetLockLevel(List *cmds)
break;
case AT_DetachPartition:
- cmd_lockmode = AccessExclusiveLock;
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ cmd_lockmode = ShareUpdateExclusiveLock;
+ else
+ cmd_lockmode = AccessExclusiveLock;
+ break;
+
+ case AT_DetachPartitionFinalize:
+ cmd_lockmode = ShareUpdateExclusiveLock;
break;
case AT_CheckNotNull:
@@ -4218,6 +4233,19 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* Find or create work queue entry for this table */
tab = ATGetQueueEntry(wqueue, rel);
+ /*
+ * Disallow any ALTER TABLE other than ALTER TABLE DETACH FINALIZE on
+ * partitions that are pending detach.
+ */
+ if (rel->rd_rel->relispartition &&
+ cmd->subtype != AT_DetachPartitionFinalize &&
+ PartitionHasPendingDetach(RelationGetRelid(rel)))
+ ereport(ERROR,
+ errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot alter partition \"%s\" with an incomplete detach",
+ RelationGetRelationName(rel)),
+ errhint("Use ALTER TABLE ... DETACH PARTITION ... FINALIZE to complete the pending detach operation."));
+
/*
* Copy the original subcommand for each table. This avoids conflicts
* when different child tables need to make different parse
@@ -4531,6 +4559,11 @@ ATPrepCmd(List **wqueue, Relation rel, AlterTableCmd *cmd,
/* No command-specific prep needed */
pass = AT_PASS_MISC;
break;
+ case AT_DetachPartitionFinalize:
+ ATSimplePermissions(rel, ATT_TABLE);
+ /* No command-specific prep needed */
+ pass = AT_PASS_MISC;
+ break;
default: /* oops */
elog(ERROR, "unrecognized alter table type: %d",
(int) cmd->subtype);
@@ -4920,7 +4953,12 @@ ATExecCmd(List **wqueue, AlteredTableInfo *tab,
Assert(cmd != NULL);
/* ATPrepCmd ensures it must be a table */
Assert(rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
- ATExecDetachPartition(rel, ((PartitionCmd *) cmd->def)->name);
+ ATExecDetachPartition(wqueue, tab, rel,
+ ((PartitionCmd *) cmd->def)->name,
+ ((PartitionCmd *) cmd->def)->concurrent);
+ break;
+ case AT_DetachPartitionFinalize:
+ ATExecDetachPartitionFinalize(rel, ((PartitionCmd *) cmd->def)->name);
break;
case AT_AlterCollationRefreshVersion:
/* ATPrepCmd ensured it must be an index */
@@ -6352,7 +6390,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
*/
if (colDef->identity &&
recurse &&
- find_inheritance_children(myrelid, NoLock) != NIL)
+ find_inheritance_children(myrelid, false, NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("cannot recursively add identity column to table that has child tables")));
@@ -6597,7 +6635,8 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* If we are told not to recurse, there had better not be any child
@@ -6751,7 +6790,7 @@ ATPrepDropNotNull(Relation rel, bool recurse, bool recursing)
*/
if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
Assert(partdesc != NULL);
if (partdesc->nparts > 0 && !recurse && !recursing)
@@ -6850,7 +6889,7 @@ ATExecDropNotNull(Relation rel, const char *colName, LOCKMODE lockmode)
/* If rel is partition, shouldn't drop NOT NULL if parent has the same */
if (rel->rd_rel->relispartition)
{
- Oid parentId = get_partition_parent(RelationGetRelid(rel));
+ Oid parentId = get_partition_parent(RelationGetRelid(rel), false);
Relation parent = table_open(parentId, AccessShareLock);
TupleDesc tupDesc = RelationGetDescr(parent);
AttrNumber parent_attnum;
@@ -7460,7 +7499,7 @@ ATPrepDropExpression(Relation rel, AlterTableCmd *cmd, bool recurse, bool recurs
* resulting state can be properly dumped and restored.
*/
if (!recurse &&
- find_inheritance_children(RelationGetRelid(rel), lockmode))
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("ALTER TABLE / DROP EXPRESSION must be applied to child tables too")));
@@ -8067,7 +8106,8 @@ ATExecDropColumn(List **wqueue, Relation rel, const char *colName,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
if (children)
{
@@ -8531,7 +8571,8 @@ ATAddCheckConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel,
* routines, we have to do this one level of recursion at a time; we can't
* use find_all_inheritors to do it in one pass.
*/
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
/*
* Check if ONLY was specified with ALTER TABLE. If so, allow the
@@ -9146,7 +9187,7 @@ addFkRecurseReferenced(List **wqueue, Constraint *fkconstraint, Relation rel,
*/
if (pkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(pkrel);
+ PartitionDesc pd = RelationGetPartitionDesc(pkrel, false);
for (int i = 0; i < pd->nparts; i++)
{
@@ -9280,7 +9321,7 @@ addFkRecurseReferencing(List **wqueue, Constraint *fkconstraint, Relation rel,
}
else if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc pd = RelationGetPartitionDesc(rel);
+ PartitionDesc pd = RelationGetPartitionDesc(rel, false);
/*
* Recurse to take appropriate action on each partition; either we
@@ -11064,7 +11105,8 @@ ATExecDropConstraint(Relation rel, const char *constrName,
* use find_all_inheritors to do it in one pass.
*/
if (!is_no_inherit_constraint)
- children = find_inheritance_children(RelationGetRelid(rel), lockmode);
+ children =
+ find_inheritance_children(RelationGetRelid(rel), false, lockmode);
else
children = NIL;
@@ -11448,7 +11490,8 @@ ATPrepAlterColumnType(List **wqueue,
}
}
else if (!recursing &&
- find_inheritance_children(RelationGetRelid(rel), NoLock) != NIL)
+ find_inheritance_children(RelationGetRelid(rel), false,
+ NoLock) != NIL)
ereport(ERROR,
(errcode(ERRCODE_INVALID_TABLE_DEFINITION),
errmsg("type of inherited column \"%s\" must be changed in child tables too",
@@ -14286,7 +14329,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
*/
/* Off to RemoveInheritance() where most of the work happens */
- RemoveInheritance(rel, parent_rel);
+ RemoveInheritance(rel, parent_rel, false);
ObjectAddressSet(address, RelationRelationId,
RelationGetRelid(parent_rel));
@@ -14297,12 +14340,72 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
return address;
}
+/*
+ * MarkInheritDetached
+ *
+ * Set inhdetachpending for a partition, for ATExecDetachPartition
+ * in concurrent mode.
+ */
+static void
+MarkInheritDetached(Relation child_rel, Relation parent_rel)
+{
+ Relation catalogRelation;
+ SysScanDesc scan;
+ ScanKeyData key;
+ HeapTuple inheritsTuple;
+ bool found = false;
+
+ Assert(child_rel->rd_rel->relkind == RELKIND_RELATION ||
+ child_rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
+ Assert(parent_rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE);
+
+ /*
+ * Find pg_inherits entries by inhrelid.
+ */
+ catalogRelation = table_open(InheritsRelationId, RowExclusiveLock);
+ ScanKeyInit(&key,
+ Anum_pg_inherits_inhrelid,
+ BTEqualStrategyNumber, F_OIDEQ,
+ ObjectIdGetDatum(RelationGetRelid(child_rel)));
+ scan = systable_beginscan(catalogRelation, InheritsRelidSeqnoIndexId,
+ true, NULL, 1, &key);
+
+ while (HeapTupleIsValid(inheritsTuple = systable_getnext(scan)))
+ {
+ HeapTuple newtup;
+
+ if (((Form_pg_inherits) GETSTRUCT(inheritsTuple))->inhparent !=
+ RelationGetRelid(parent_rel))
+ elog(ERROR, "bad parent tuple found for partition %u",
+ RelationGetRelid(child_rel));
+
+ newtup = heap_copytuple(inheritsTuple);
+ ((Form_pg_inherits) GETSTRUCT(newtup))->inhdetachpending = true;
+
+ CatalogTupleUpdate(catalogRelation,
+ &inheritsTuple->t_self,
+ newtup);
+ found = true;
+ }
+
+ /* Done */
+ systable_endscan(scan);
+ table_close(catalogRelation, RowExclusiveLock);
+
+ if (!found)
+ ereport(ERROR,
+ (errcode(ERRCODE_UNDEFINED_TABLE),
+ errmsg("relation \"%s\" is not a partition of relation \"%s\"",
+ RelationGetRelationName(child_rel),
+ RelationGetRelationName(parent_rel))));
+}
+
/*
* RemoveInheritance
*
* Drop a parent from the child's parents. This just adjusts the attinhcount
* and attislocal of the columns and removes the pg_inherit and pg_depend
- * entries.
+ * entries. expect_detached is passed down to DeleteInheritsTuple, q.v..
*
* If attinhcount goes to 0 then attislocal gets set to true. If it goes back
* up attislocal stays true, which means if a child is ever removed from a
@@ -14316,7 +14419,7 @@ ATExecDropInherit(Relation rel, RangeVar *parent, LOCKMODE lockmode)
* Common to ATExecDropInherit() and ATExecDetachPartition().
*/
static void
-RemoveInheritance(Relation child_rel, Relation parent_rel)
+RemoveInheritance(Relation child_rel, Relation parent_rel, bool expect_detached)
{
Relation catalogRelation;
SysScanDesc scan;
@@ -14332,7 +14435,9 @@ RemoveInheritance(Relation child_rel, Relation parent_rel)
child_is_partition = true;
found = DeleteInheritsTuple(RelationGetRelid(child_rel),
- RelationGetRelid(parent_rel));
+ RelationGetRelid(parent_rel),
+ expect_detached,
+ RelationGetRelationName(child_rel));
if (!found)
{
if (child_is_partition)
@@ -16498,7 +16603,7 @@ QueuePartitionConstraintValidation(List **wqueue, Relation scanrel,
}
else if (scanrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(scanrel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(scanrel, false);
int i;
for (i = 0; i < partdesc->nparts; i++)
@@ -16558,7 +16663,7 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd,
* new partition will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
LockRelationOid(defaultPartOid, AccessExclusiveLock);
@@ -17147,105 +17252,229 @@ CloneRowTriggersToPartition(Relation parent, Relation partition)
* ALTER TABLE DETACH PARTITION
*
* Return the address of the relation that is no longer a partition of rel.
+ *
+ * If concurrent mode is requested, we run in two transactions. A side-
+ * effect is that this command cannot run in a multi-part ALTER TABLE.
+ * Currently, that's enforced by the grammar.
+ *
+ * The strategy for concurrency is to first modify the partition's
+ * pg_inherit catalog row to make it visible to everyone that the
+ * partition is detached, lock the partition against writes, and commit
+ * the transaction; anyone who requests the partition descriptor from
+ * that point onwards has to ignore such a partition. In a second
+ * transaction, we wait until all transactions that could have seen the
+ * partition as attached are gone, then we remove the rest of partition
+ * metadata (pg_inherits and pg_class.relpartbounds).
*/
static ObjectAddress
-ATExecDetachPartition(Relation rel, RangeVar *name)
+ATExecDetachPartition(List **wqueue, AlteredTableInfo *tab, Relation rel,
+ RangeVar *name, bool concurrent)
{
- Relation partRel,
- classRel;
- HeapTuple tuple,
- newtuple;
- Datum new_val[Natts_pg_class];
- bool new_null[Natts_pg_class],
- new_repl[Natts_pg_class];
+ Relation partRel;
ObjectAddress address;
Oid defaultPartOid;
- List *indexes;
- List *fks;
- ListCell *cell;
/*
* We must lock the default partition, because detaching this partition
* will change its partition constraint.
*/
defaultPartOid =
- get_default_oid_from_partdesc(RelationGetPartitionDesc(rel));
- if (OidIsValid(defaultPartOid))
- LockRelationOid(defaultPartOid, AccessExclusiveLock);
-
- partRel = table_openrv(name, ShareUpdateExclusiveLock);
-
- /* Ensure that foreign keys still hold after this detach */
- ATDetachCheckNoForeignKeyRefs(partRel);
-
- /* All inheritance related checks are performed within the function */
- RemoveInheritance(partRel, rel);
-
- /* Update pg_class tuple */
- classRel = table_open(RelationRelationId, RowExclusiveLock);
- tuple = SearchSysCacheCopy1(RELOID,
- ObjectIdGetDatum(RelationGetRelid(partRel)));
- if (!HeapTupleIsValid(tuple))
- elog(ERROR, "cache lookup failed for relation %u",
- RelationGetRelid(partRel));
- Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
-
- /* Clear relpartbound and reset relispartition */
- memset(new_val, 0, sizeof(new_val));
- memset(new_null, false, sizeof(new_null));
- memset(new_repl, false, sizeof(new_repl));
- new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
- new_null[Anum_pg_class_relpartbound - 1] = true;
- new_repl[Anum_pg_class_relpartbound - 1] = true;
- newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
- new_val, new_null, new_repl);
-
- ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
- CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
- heap_freetuple(newtuple);
-
+ get_default_oid_from_partdesc(RelationGetPartitionDesc(rel, false));
if (OidIsValid(defaultPartOid))
{
/*
- * If the relation being detached is the default partition itself,
- * remove it from the parent's pg_partitioned_table entry.
+ * Concurrent detaching when a default partition exists is not
+ * supported. The main problem is that the default partition
+ * constraint would change. And there's a definitional problem: what
+ * should happen to the tuples that are being inserted that belong to
+ * the partition being detached? Putting them on the partition being
+ * detached would be wrong, since they'd become "lost" after the but
+ * we cannot put them in the default partition either until we alter
+ * its partition constraint.
*
- * If not, we must invalidate default partition's relcache entry, as
- * in StorePartitionBound: its partition constraint depends on every
- * other partition's partition constraint.
+ * I think we could solve this problem if we effected the constraint
+ * change before committing the first transaction. But the lock would
+ * have to remain AEL and it would cause concurrent query planning to
+ * be blocked, so changing it that way would be even worse.
*/
- if (RelationGetRelid(partRel) == defaultPartOid)
- update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
- else
- CacheInvalidateRelcacheByRelid(defaultPartOid);
+ if (concurrent)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("cannot detach partitions concurrently when a default partition exists")));
+ LockRelationOid(defaultPartOid, AccessExclusiveLock);
}
- /* detach indexes too */
- indexes = RelationGetIndexList(partRel);
- foreach(cell, indexes)
+ /*
+ * In concurrent mode, the partition is locked with share-update-exclusive
+ * in the first transaction. This allows concurrent transactions to be
+ * doing DML to the partition.
+ */
+ partRel = table_openrv(name, concurrent ? ShareUpdateExclusiveLock :
+ AccessExclusiveLock);
+
+ /*
+ * Check inheritance conditions and either delete the pg_inherits row
+ * (in non-concurrent mode) or just set the inhdetachpending flag.
+ */
+ if (!concurrent)
+ RemoveInheritance(partRel, rel, false);
+ else
+ MarkInheritDetached(partRel, rel);
+
+ /*
+ * Ensure that foreign keys still hold after this detach. This keeps
+ * locks on the referencing tables, which prevents concurrent transactions
+ * from adding rows that we wouldn't see. For this to work in concurrent
+ * mode, it is critical that the partition appears as no longer attached
+ * for the RI queries as soon as the first transaction commits.
+ */
+ ATDetachCheckNoForeignKeyRefs(partRel);
+
+ /*
+ * Concurrent mode has to work harder; first we add a new constraint to
+ * the partition that matches the partition constraint. Then we close our
+ * existing transaction, and in a new one wait for all processes to catch
+ * up on the catalog updates we've done so far; at that point we can
+ * complete the operation.
+ */
+ if (concurrent)
{
- Oid idxid = lfirst_oid(cell);
- Relation idx;
- Oid constrOid;
+ Oid partrelid,
+ parentrelid;
+ LOCKTAG tag;
+ char *parentrelname;
+ char *partrelname;
- if (!has_superclass(idxid))
- continue;
+ /*
+ * Add a new constraint to the partition being detached, which
+ * supplants the partition constraint (unless there is one already).
+ */
+ DetachAddConstraintIfNeeded(wqueue, partRel);
- Assert((IndexGetRelation(get_partition_parent(idxid), false) ==
- RelationGetRelid(rel)));
+ /*
+ * We're almost done now; the only traces that remain are the
+ * pg_inherits tuple and the partition's relpartbounds. Before we can
+ * remove those, we need to wait until all transactions that know that
+ * this is a partition are gone.
+ */
- idx = index_open(idxid, AccessExclusiveLock);
- IndexSetParentIndex(idx, InvalidOid);
+ /*
+ * Remember relation OIDs to re-acquire them later; and relation names
+ * too, for error messages if something is dropped in between.
+ */
+ partrelid = RelationGetRelid(partRel);
+ parentrelid = RelationGetRelid(rel);
+ parentrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(rel));
+ partrelname = MemoryContextStrdup(PortalContext,
+ RelationGetRelationName(partRel));
- /* If there's a constraint associated with the index, detach it too */
- constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
- idxid);
- if (OidIsValid(constrOid))
- ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+ /* Invalidate relcache entries for the parent -- must be before close */
+ CacheInvalidateRelcache(rel);
+
+ table_close(partRel, NoLock);
+ table_close(rel, NoLock);
+ tab->rel = NULL;
+
+ /* Make updated catalog entry visible */
+ PopActiveSnapshot();
+ CommitTransactionCommand();
+
+ StartTransactionCommand();
+
+ /*
+ * Now wait. This ensures that all queries that were planned including
+ * the partition are finished before we remove the rest of catalog
+ * entries. We don't need or indeed want to acquire this lock, though
+ * -- that would block later queries.
+ *
+ * We don't need to concern ourselves with waiting for a lock on the
+ * partition itself, since we will acquire AccessExclusiveLock below.
+ */
+ SET_LOCKTAG_RELATION(tag, MyDatabaseId, parentrelid);
+ WaitForLockersMultiple(list_make1(&tag), AccessExclusiveLock, false);
+
+ /*
+ * Now acquire locks in both relations again. Note they may have been
+ * removed in the meantime, so care is required.
+ */
+ rel = try_relation_open(parentrelid, ShareUpdateExclusiveLock);
+ partRel = try_relation_open(partrelid, AccessExclusiveLock);
+
+ /* If the relations aren't there, something bad happened; bail out */
+ if (rel == NULL)
+ {
+ if (partRel != NULL) /* shouldn't happen */
+ elog(WARNING, "dangling partition \"%s\" remains, can't fix",
+ partrelname);
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partitioned table \"%s\" was removed concurrently",
+ parentrelname)));
+ }
+ if (partRel == NULL)
+ ereport(ERROR,
+ (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
+ errmsg("partition \"%s\" was removed concurrently", partrelname)));
+
+ tab->rel = rel;
- index_close(idx, NoLock);
}
- table_close(classRel, RowExclusiveLock);
+
+ /* Do the final part of detaching */
+ DetachPartitionFinalize(rel, partRel, concurrent, defaultPartOid, false);
+
+ ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
+
+ /* keep our lock until commit */
+ table_close(partRel, NoLock);
+
+ return address;
+}
+
+/*
+ * Second part of ALTER TABLE .. DETACH.
+ *
+ * This is separate so that it can be run independently when the second
+ * transaction of the concurrent algorithm fails (crash or abort).
+ */
+static void
+DetachPartitionFinalize(Relation rel, Relation partRel, bool concurrent,
+ Oid defaultPartOid, bool wait)
+{
+ Relation classRel;
+ List *fks;
+ ListCell *cell;
+ List *indexes;
+ Datum new_val[Natts_pg_class];
+ bool new_null[Natts_pg_class],
+ new_repl[Natts_pg_class];
+ HeapTuple tuple,
+ newtuple;
+
+ /*
+ * If asked to, wait until existing snapshots are gone. This is important
+ * if the second transaction of DETACH PARTITION CONCURRENTLY is canceled:
+ * the user could immediately run DETACH FINALIZE without actually waiting
+ * for existing transactions. We must not complete the detach action until
+ * all such queries are complete (otherwise we would present them with an
+ * inconsistent view of catalogs).
+ */
+ if (wait)
+ {
+ Snapshot snap = GetActiveSnapshot();
+
+ WaitForOlderSnapshots(snap->xmin, false);
+ }
+
+ if (concurrent)
+ {
+ /*
+ * We can remove the pg_inherits row now. (In the non-concurrent case,
+ * this was already done).
+ */
+ RemoveInheritance(partRel, rel, true);
+ }
/* Drop any triggers that were cloned on creation/attach. */
DropClonedTriggersFromPartition(RelationGetRelid(partRel));
@@ -17318,22 +17547,150 @@ ATExecDetachPartition(Relation rel, RangeVar *name)
ObjectAddressSet(constraint, ConstraintRelationId, constrOid);
performDeletion(&constraint, DROP_RESTRICT, 0);
}
- CommandCounterIncrement();
+
+ /* Now we can detach indexes */
+ indexes = RelationGetIndexList(partRel);
+ foreach(cell, indexes)
+ {
+ Oid idxid = lfirst_oid(cell);
+ Relation idx;
+ Oid constrOid;
+
+ if (!has_superclass(idxid))
+ continue;
+
+ Assert((IndexGetRelation(get_partition_parent(idxid, false), false) ==
+ RelationGetRelid(rel)));
+
+ idx = index_open(idxid, AccessExclusiveLock);
+ IndexSetParentIndex(idx, InvalidOid);
+
+ /* If there's a constraint associated with the index, detach it too */
+ constrOid = get_relation_idx_constraint_oid(RelationGetRelid(partRel),
+ idxid);
+ if (OidIsValid(constrOid))
+ ConstraintSetParentConstraint(constrOid, InvalidOid, InvalidOid);
+
+ index_close(idx, NoLock);
+ }
+
+ /* Update pg_class tuple */
+ classRel = table_open(RelationRelationId, RowExclusiveLock);
+ tuple = SearchSysCacheCopy1(RELOID,
+ ObjectIdGetDatum(RelationGetRelid(partRel)));
+ if (!HeapTupleIsValid(tuple))
+ elog(ERROR, "cache lookup failed for relation %u",
+ RelationGetRelid(partRel));
+ Assert(((Form_pg_class) GETSTRUCT(tuple))->relispartition);
+
+ /* Clear relpartbound and reset relispartition */
+ memset(new_val, 0, sizeof(new_val));
+ memset(new_null, false, sizeof(new_null));
+ memset(new_repl, false, sizeof(new_repl));
+ new_val[Anum_pg_class_relpartbound - 1] = (Datum) 0;
+ new_null[Anum_pg_class_relpartbound - 1] = true;
+ new_repl[Anum_pg_class_relpartbound - 1] = true;
+ newtuple = heap_modify_tuple(tuple, RelationGetDescr(classRel),
+ new_val, new_null, new_repl);
+
+ ((Form_pg_class) GETSTRUCT(newtuple))->relispartition = false;
+ CatalogTupleUpdate(classRel, &newtuple->t_self, newtuple);
+ heap_freetuple(newtuple);
+ table_close(classRel, RowExclusiveLock);
+
+ if (OidIsValid(defaultPartOid))
+ {
+ /*
+ * If the relation being detached is the default partition itself,
+ * remove it from the parent's pg_partitioned_table entry.
+ *
+ * If not, we must invalidate default partition's relcache entry, as
+ * in StorePartitionBound: its partition constraint depends on every
+ * other partition's partition constraint.
+ */
+ if (RelationGetRelid(partRel) == defaultPartOid)
+ update_default_partition_oid(RelationGetRelid(rel), InvalidOid);
+ else
+ CacheInvalidateRelcacheByRelid(defaultPartOid);
+ }
/*
* Invalidate the parent's relcache so that the partition is no longer
* included in its partition descriptor.
*/
CacheInvalidateRelcache(rel);
+}
+
+/*
+ * ALTER TABLE ... DETACH PARTITION ... FINALIZE
+ *
+ * To use when a DETACH PARTITION command previously did not run to
+ * completion; this completes the detaching process.
+ */
+static ObjectAddress
+ATExecDetachPartitionFinalize(Relation rel, RangeVar *name)
+{
+ Relation partRel;
+ ObjectAddress address;
+
+ partRel = table_openrv(name, AccessExclusiveLock);
+
+ DetachPartitionFinalize(rel, partRel, true, InvalidOid, true);
ObjectAddressSet(address, RelationRelationId, RelationGetRelid(partRel));
- /* keep our lock until commit */
table_close(partRel, NoLock);
return address;
}
+/*
+ * DetachAddConstraintIfNeeded
+ * Subroutine for ATExecDetachPartition. Create a constraint that
+ * takes the place of the partition constraint, but avoid creating
+ * a dupe if an equivalent constraint already exists.
+ */
+static void
+DetachAddConstraintIfNeeded(List **wqueue, Relation partRel)
+{
+ AlteredTableInfo *tab;
+ Expr *constraintExpr;
+ TupleDesc td = RelationGetDescr(partRel);
+ Constraint *n;
+
+ constraintExpr = make_ands_explicit(RelationGetPartitionQual(partRel));
+
+ /* If an identical constraint exists, we don't need to create one */
+ if (td->constr && td->constr->num_check > 0)
+ {
+ for (int i = 0; i < td->constr->num_check; i++)
+ {
+ Node *thisconstr;
+
+ thisconstr = stringToNode(td->constr->check[i].ccbin);
+
+ if (equal(constraintExpr, thisconstr))
+ return;
+ }
+ }
+
+ tab = ATGetQueueEntry(wqueue, partRel);
+
+ /* Add constraint on partition, equivalent to the partition constraint */
+ n = makeNode(Constraint);
+ n->contype = CONSTR_CHECK;
+ n->conname = NULL;
+ n->location = -1;
+ n->is_no_inherit = false;
+ n->raw_expr = NULL;
+ n->cooked_expr = nodeToString(constraintExpr);
+ n->initially_valid = true;
+ n->skip_validation = true;
+ /* It's a re-add, since it nominally already exists */
+ ATAddCheckConstraint(wqueue, tab, partRel, n,
+ true, false, true, ShareUpdateExclusiveLock);
+}
+
/*
* DropClonedTriggersFromPartition
* subroutine for ATExecDetachPartition to remove any triggers that were
@@ -17501,7 +17858,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
/* Silently do nothing if already in the right state */
currParent = partIdx->rd_rel->relispartition ?
- get_partition_parent(partIdxId) : InvalidOid;
+ get_partition_parent(partIdxId, false) : InvalidOid;
if (currParent != RelationGetRelid(parentIdx))
{
IndexInfo *childInfo;
@@ -17529,7 +17886,7 @@ ATExecAttachPartitionIdx(List **wqueue, Relation parentIdx, RangeVar *name)
RelationGetRelationName(partIdx))));
/* Make sure it indexes a partition of the other index's table */
- partDesc = RelationGetPartitionDesc(parentTbl);
+ partDesc = RelationGetPartitionDesc(parentTbl, false);
found = false;
for (i = 0; i < partDesc->nparts; i++)
{
@@ -17683,7 +18040,7 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
* If we found as many inherited indexes as the partitioned table has
* partitions, we're good; update pg_index to set indisvalid.
*/
- if (tuples == RelationGetPartitionDesc(partedTbl)->nparts)
+ if (tuples == RelationGetPartitionDesc(partedTbl, false)->nparts)
{
Relation idxRel;
HeapTuple newtup;
@@ -17713,8 +18070,8 @@ validatePartitionedIndex(Relation partedIdx, Relation partedTbl)
/* make sure we see the validation we just did */
CommandCounterIncrement();
- parentIdxId = get_partition_parent(RelationGetRelid(partedIdx));
- parentTblId = get_partition_parent(RelationGetRelid(partedTbl));
+ parentIdxId = get_partition_parent(RelationGetRelid(partedIdx), false);
+ parentTblId = get_partition_parent(RelationGetRelid(partedTbl), false);
parentIdx = relation_open(parentIdxId, AccessExclusiveLock);
parentTbl = relation_open(parentTblId, AccessExclusiveLock);
Assert(!parentIdx->rd_index->indisvalid);
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 4e4e05844c..7383d5994e 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -1119,7 +1119,7 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
*/
if (partition_recurse)
{
- PartitionDesc partdesc = RelationGetPartitionDesc(rel);
+ PartitionDesc partdesc = RelationGetPartitionDesc(rel, false);
List *idxs = NIL;
List *childTbls = NIL;
ListCell *l;
@@ -1141,7 +1141,8 @@ CreateTrigger(CreateTrigStmt *stmt, const char *queryString,
ListCell *l;
List *idxs = NIL;
- idxs = find_inheritance_children(indexOid, ShareRowExclusiveLock);
+ idxs = find_inheritance_children(indexOid, false,
+ ShareRowExclusiveLock);
foreach(l, idxs)
childTbls = lappend_oid(childTbls,
IndexGetRelation(lfirst_oid(l),
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index b8da4c5967..619aaffae4 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -569,6 +569,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
int partidx)
{
ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+ Oid partOid = dispatch->partdesc->oids[partidx];
Relation partrel;
int firstVarno = mtstate->resultRelInfo[0].ri_RangeTableIndex;
Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
@@ -579,7 +580,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
oldcxt = MemoryContextSwitchTo(proute->memcxt);
- partrel = table_open(dispatch->partdesc->oids[partidx], RowExclusiveLock);
+ partrel = table_open(partOid, RowExclusiveLock);
leaf_part_rri = makeNode(ResultRelInfo);
InitResultRelInfo(leaf_part_rri,
@@ -1065,9 +1066,21 @@ ExecInitPartitionDispatchInfo(EState *estate,
int dispatchidx;
MemoryContext oldcxt;
+ /*
+ * For data modification, it is better that executor does not include
+ * partitions being detached, except in snapshot-isolation mode. This
+ * means that a read-committed transaction immediately gets a "no
+ * partition for tuple" error when a tuple is inserted into a partition
+ * that's being detached concurrently, but a transaction in repeatable-
+ * read mode can still use the partition. Note that because partition
+ * detach uses ShareLock on the partition (which conflicts with DML),
+ * we're certain that the detach won't be able to complete until any
+ * inserting transaction is done.
+ */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt,
+ IsolationUsesXactSnapshot());
oldcxt = MemoryContextSwitchTo(proute->memcxt);
@@ -1645,9 +1658,10 @@ ExecCreatePartitionPruneState(PlanState *planstate,
ListCell *lc;
int i;
+ /* Executor must always include detached partitions */
if (estate->es_partition_directory == NULL)
estate->es_partition_directory =
- CreatePartitionDirectory(estate->es_query_cxt);
+ CreatePartitionDirectory(estate->es_query_cxt, true);
n_part_hierarchies = list_length(partitionpruneinfo->prune_infos);
Assert(n_part_hierarchies > 0);
@@ -1713,9 +1727,12 @@ ExecCreatePartitionPruneState(PlanState *planstate,
partrel);
/*
- * Initialize the subplan_map and subpart_map. Since detaching a
- * partition requires AccessExclusiveLock, no partitions can have
- * disappeared, nor can the bounds for any partition have changed.
+ * Initialize the subplan_map and subpart_map.
+ *
+ * Because we request detached partitions to be included, and
+ * detaching waits for old transactions, it is safe to assume that
+ * no partitions have disappeared since this query was planned.
+ *
* However, new partitions may have been added.
*/
Assert(partdesc->nparts >= pinfo->nparts);
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 82d7cce5d5..38b56231b7 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -4737,6 +4737,7 @@ _copyPartitionCmd(const PartitionCmd *from)
COPY_NODE_FIELD(name);
COPY_NODE_FIELD(bound);
+ COPY_SCALAR_FIELD(concurrent);
return newnode;
}
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index 3e980c457c..3292dda342 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2975,6 +2975,7 @@ _equalPartitionCmd(const PartitionCmd *a, const PartitionCmd *b)
{
COMPARE_NODE_FIELD(name);
COMPARE_NODE_FIELD(bound);
+ COMPARE_SCALAR_FIELD(concurrent);
return true;
}
diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c
index 7f2e40ae39..6c39bf893f 100644
--- a/src/backend/optimizer/util/plancat.c
+++ b/src/backend/optimizer/util/plancat.c
@@ -2141,10 +2141,14 @@ set_relation_partition_info(PlannerInfo *root, RelOptInfo *rel,
{
PartitionDesc partdesc;
- /* Create the PartitionDirectory infrastructure if we didn't already */
+ /*
+ * Create the PartitionDirectory infrastructure if we didn't already.
+ */
if (root->glob->partition_directory == NULL)
+ {
root->glob->partition_directory =
- CreatePartitionDirectory(CurrentMemoryContext);
+ CreatePartitionDirectory(CurrentMemoryContext, false);
+ }
partdesc = PartitionDirectoryLookup(root->glob->partition_directory,
relation);
diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y
index 4843ff99e2..2132cf4d82 100644
--- a/src/backend/parser/gram.y
+++ b/src/backend/parser/gram.y
@@ -658,7 +658,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query);
EXCLUDE EXCLUDING EXCLUSIVE EXECUTE EXISTS EXPLAIN EXPRESSION
EXTENSION EXTERNAL EXTRACT
- FALSE_P FAMILY FETCH FILTER FIRST_P FLOAT_P FOLLOWING FOR
+ FALSE_P FAMILY FETCH FILTER FINALIZE FIRST_P FLOAT_P FOLLOWING FOR
FORCE FOREIGN FORWARD FREEZE FROM FULL FUNCTION FUNCTIONS
GENERATED GLOBAL GRANT GRANTED GREATEST GROUP_P GROUPING GROUPS
@@ -2108,12 +2108,13 @@ partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = $4;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
}
- /* ALTER TABLE <name> DETACH PARTITION <partition_name> */
- | DETACH PARTITION qualified_name
+ /* ALTER TABLE <name> DETACH PARTITION <partition_name> [CONCURRENTLY] */
+ | DETACH PARTITION qualified_name opt_concurrently
{
AlterTableCmd *n = makeNode(AlterTableCmd);
PartitionCmd *cmd = makeNode(PartitionCmd);
@@ -2121,8 +2122,21 @@ partition_cmd:
n->subtype = AT_DetachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = $4;
n->def = (Node *) cmd;
+ $$ = (Node *) n;
+ }
+ | DETACH PARTITION qualified_name FINALIZE
+ {
+ AlterTableCmd *n = makeNode(AlterTableCmd);
+ PartitionCmd *cmd = makeNode(PartitionCmd);
+
+ n->subtype = AT_DetachPartitionFinalize;
+ cmd->name = $3;
+ cmd->bound = NULL;
+ cmd->concurrent = false;
+ n->def = (Node *) cmd;
$$ = (Node *) n;
}
;
@@ -2137,6 +2151,7 @@ index_partition_cmd:
n->subtype = AT_AttachPartition;
cmd->name = $3;
cmd->bound = NULL;
+ cmd->concurrent = false;
n->def = (Node *) cmd;
$$ = (Node *) n;
@@ -15395,6 +15410,7 @@ unreserved_keyword:
| EXTERNAL
| FAMILY
| FILTER
+ | FINALIZE
| FIRST_P
| FOLLOWING
| FORCE
@@ -15936,6 +15952,7 @@ bare_label_keyword:
| EXTRACT
| FALSE_P
| FAMILY
+ | FINALIZE
| FIRST_P
| FLOAT_P
| FOLLOWING
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index e5f3482d52..2b2b1dc1ad 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -2798,7 +2798,7 @@ check_new_partition_bound(char *relname, Relation parent,
PartitionBoundSpec *spec, ParseState *pstate)
{
PartitionKey key = RelationGetPartitionKey(parent);
- PartitionDesc partdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc partdesc = RelationGetPartitionDesc(parent, true);
PartitionBoundInfo boundinfo = partdesc->boundinfo;
int with = -1;
bool overlap = false;
@@ -3990,7 +3990,7 @@ get_qual_for_list(Relation parent, PartitionBoundSpec *spec)
{
int i;
int ndatums = 0;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
PartitionBoundInfo boundinfo = pdesc->boundinfo;
if (boundinfo)
@@ -4190,7 +4190,7 @@ get_qual_for_range(Relation parent, PartitionBoundSpec *spec,
if (spec->is_default)
{
List *or_expr_args = NIL;
- PartitionDesc pdesc = RelationGetPartitionDesc(parent);
+ PartitionDesc pdesc = RelationGetPartitionDesc(parent, true); /* XXX correct? */
Oid *inhoids = pdesc->oids;
int nparts = pdesc->nparts,
i;
diff --git a/src/backend/partitioning/partdesc.c b/src/backend/partitioning/partdesc.c
index f852b6e99d..58570fecfd 100644
--- a/src/backend/partitioning/partdesc.c
+++ b/src/backend/partitioning/partdesc.c
@@ -37,6 +37,7 @@ typedef struct PartitionDirectoryData
{
MemoryContext pdir_mcxt;
HTAB *pdir_hash;
+ bool include_detached;
} PartitionDirectoryData;
typedef struct PartitionDirectoryEntry
@@ -46,7 +47,7 @@ typedef struct PartitionDirectoryEntry
PartitionDesc pd;
} PartitionDirectoryEntry;
-static void RelationBuildPartitionDesc(Relation rel);
+static void RelationBuildPartitionDesc(Relation rel, bool include_detached);
/*
@@ -61,13 +62,14 @@ static void RelationBuildPartitionDesc(Relation rel);
* that the data doesn't become stale.
*/
PartitionDesc
-RelationGetPartitionDesc(Relation rel)
+RelationGetPartitionDesc(Relation rel, bool include_detached)
{
if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
return NULL;
- if (unlikely(rel->rd_partdesc == NULL))
- RelationBuildPartitionDesc(rel);
+ if (unlikely(rel->rd_partdesc == NULL ||
+ rel->rd_partdesc->includes_detached != include_detached))
+ RelationBuildPartitionDesc(rel, include_detached);
return rel->rd_partdesc;
}
@@ -88,7 +90,7 @@ RelationGetPartitionDesc(Relation rel)
* permanently.
*/
static void
-RelationBuildPartitionDesc(Relation rel)
+RelationBuildPartitionDesc(Relation rel, bool include_detached)
{
PartitionDesc partdesc;
PartitionBoundInfo boundinfo = NULL;
@@ -110,7 +112,8 @@ RelationBuildPartitionDesc(Relation rel)
* concurrently, whatever this function returns will be accurate as of
* some well-defined point in time.
*/
- inhoids = find_inheritance_children(RelationGetRelid(rel), NoLock);
+ inhoids = find_inheritance_children(RelationGetRelid(rel), include_detached,
+ NoLock);
nparts = list_length(inhoids);
/* Allocate working arrays for OIDs, leaf flags, and boundspecs. */
@@ -238,6 +241,7 @@ RelationBuildPartitionDesc(Relation rel)
partdesc->boundinfo = partition_bounds_copy(boundinfo, key);
partdesc->oids = (Oid *) palloc(nparts * sizeof(Oid));
partdesc->is_leaf = (bool *) palloc(nparts * sizeof(bool));
+ partdesc->includes_detached = include_detached;
/*
* Assign OIDs from the original array into mapped indexes of the
@@ -280,7 +284,7 @@ RelationBuildPartitionDesc(Relation rel)
* Create a new partition directory object.
*/
PartitionDirectory
-CreatePartitionDirectory(MemoryContext mcxt)
+CreatePartitionDirectory(MemoryContext mcxt, bool include_detached)
{
MemoryContext oldcontext = MemoryContextSwitchTo(mcxt);
PartitionDirectory pdir;
@@ -295,6 +299,7 @@ CreatePartitionDirectory(MemoryContext mcxt)
pdir->pdir_hash = hash_create("partition directory", 256, &ctl,
HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
+ pdir->include_detached = include_detached;
MemoryContextSwitchTo(oldcontext);
return pdir;
@@ -327,7 +332,7 @@ PartitionDirectoryLookup(PartitionDirectory pdir, Relation rel)
*/
RelationIncrementReferenceCount(rel);
pde->rel = rel;
- pde->pd = RelationGetPartitionDesc(rel);
+ pde->pd = RelationGetPartitionDesc(rel, pdir->include_detached);
Assert(pde->pd != NULL);
}
return pde->pd;
diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c
index 05bb698cf4..729274b330 100644
--- a/src/backend/tcop/utility.c
+++ b/src/backend/tcop/utility.c
@@ -1236,6 +1236,25 @@ ProcessUtilitySlow(ParseState *pstate,
AlterTableStmt *atstmt = (AlterTableStmt *) parsetree;
Oid relid;
LOCKMODE lockmode;
+ ListCell *cell;
+
+ /*
+ * Disallow ALTER TABLE .. DETACH CONCURRENTLY in a
+ * transaction block or function. (Perhaps it could be
+ * allowed in a procedure, but don't hold your breath.)
+ */
+ foreach(cell, atstmt->cmds)
+ {
+ AlterTableCmd *cmd = (AlterTableCmd *) lfirst(cell);
+
+ /* Disallow DETACH CONCURRENTLY in a transaction block */
+ if (cmd->subtype == AT_DetachPartition)
+ {
+ if (((PartitionCmd *) cmd->def)->concurrent)
+ PreventInTransactionBlock(isTopLevel,
+ "ALTER TABLE ... DETACH CONCURRENTLY");
+ }
+ }
/*
* Figure out lock mode, and acquire lock. This also does
diff --git a/src/backend/utils/adt/ri_triggers.c b/src/backend/utils/adt/ri_triggers.c
index 09a2ad2881..7c77c338ce 100644
--- a/src/backend/utils/adt/ri_triggers.c
+++ b/src/backend/utils/adt/ri_triggers.c
@@ -392,11 +392,15 @@ RI_FKey_check(TriggerData *trigdata)
/*
* Now check that foreign key exists in PK table
+ *
+ * XXX detectNewRows must be true when a partitioned table is on the
+ * referenced side. The reason is that our snapshot must be fresh
+ * in order for the hack in find_inheritance_children() to work.
*/
ri_PerformCheck(riinfo, &qkey, qplan,
fk_rel, pk_rel,
NULL, newslot,
- false,
+ pk_rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE,
SPI_OK_SELECT);
if (SPI_finish() != SPI_OK_FINISH)
diff --git a/src/backend/utils/cache/partcache.c b/src/backend/utils/cache/partcache.c
index a6388d980e..21e60f0c5e 100644
--- a/src/backend/utils/cache/partcache.c
+++ b/src/backend/utils/cache/partcache.c
@@ -341,6 +341,7 @@ generate_partition_qual(Relation rel)
bool isnull;
List *my_qual = NIL,
*result = NIL;
+ Oid parentrelid;
Relation parent;
/* Guard against stack overflow due to overly deep partition tree */
@@ -350,9 +351,14 @@ generate_partition_qual(Relation rel)
if (rel->rd_partcheckvalid)
return copyObject(rel->rd_partcheck);
- /* Grab at least an AccessShareLock on the parent table */
- parent = relation_open(get_partition_parent(RelationGetRelid(rel)),
- AccessShareLock);
+ /*
+ * Grab at least an AccessShareLock on the parent table. Must do this
+ * even if the partition has been partially detached, because transactions
+ * concurrent with the detach might still be trying to use a partition
+ * descriptor that includes it.
+ */
+ parentrelid = get_partition_parent(RelationGetRelid(rel), true);
+ parent = relation_open(parentrelid, AccessShareLock);
/* Get pg_class.relpartbound */
tuple = SearchSysCache1(RELOID, RelationGetRelid(rel));
diff --git a/src/bin/psql/describe.c b/src/bin/psql/describe.c
index eeac0efc4f..c9f7118a5d 100644
--- a/src/bin/psql/describe.c
+++ b/src/bin/psql/describe.c
@@ -2144,7 +2144,12 @@ describeOneTableDetails(const char *schemaname,
printfPQExpBuffer(&buf,
"SELECT inhparent::pg_catalog.regclass,\n"
- " pg_catalog.pg_get_expr(c.relpartbound, c.oid)");
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid),\n ");
+
+ appendPQExpBuffer(&buf,
+ pset.sversion >= 140000 ? "inhdetachpending" :
+ "false as inhdetachpending");
+
/* If verbose, also request the partition constraint definition */
if (verbose)
appendPQExpBufferStr(&buf,
@@ -2162,17 +2167,19 @@ describeOneTableDetails(const char *schemaname,
{
char *parent_name = PQgetvalue(result, 0, 0);
char *partdef = PQgetvalue(result, 0, 1);
+ char *detached = PQgetvalue(result, 0, 2);
- printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s"), parent_name,
- partdef);
+ printfPQExpBuffer(&tmpbuf, _("Partition of: %s %s%s"), parent_name,
+ partdef,
+ strcmp(detached, "t") == 0 ? " DETACH PENDING" : "");
printTableAddFooter(&cont, tmpbuf.data);
if (verbose)
{
char *partconstraintdef = NULL;
- if (!PQgetisnull(result, 0, 2))
- partconstraintdef = PQgetvalue(result, 0, 2);
+ if (!PQgetisnull(result, 0, 3))
+ partconstraintdef = PQgetvalue(result, 0, 3);
/* If there isn't any constraint, show that explicitly */
if (partconstraintdef == NULL || partconstraintdef[0] == '\0')
printfPQExpBuffer(&tmpbuf, _("No partition constraint"));
@@ -3224,9 +3231,20 @@ describeOneTableDetails(const char *schemaname,
}
/* print child tables (with additional info if partitions) */
- if (pset.sversion >= 100000)
+ if (pset.sversion >= 140000)
printfPQExpBuffer(&buf,
"SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ " inhdetachpending,"
+ " pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
+ "FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
+ "WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
+ "ORDER BY pg_catalog.pg_get_expr(c.relpartbound, c.oid) = 'DEFAULT',"
+ " c.oid::pg_catalog.regclass::pg_catalog.text;",
+ oid);
+ else if (pset.sversion >= 100000)
+ printfPQExpBuffer(&buf,
+ "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ " false AS inhdetachpending,"
" pg_catalog.pg_get_expr(c.relpartbound, c.oid)\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
@@ -3235,14 +3253,16 @@ describeOneTableDetails(const char *schemaname,
oid);
else if (pset.sversion >= 80300)
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ " false AS inhdetachpending, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.oid::pg_catalog.regclass::pg_catalog.text;",
oid);
else
printfPQExpBuffer(&buf,
- "SELECT c.oid::pg_catalog.regclass, c.relkind, NULL\n"
+ "SELECT c.oid::pg_catalog.regclass, c.relkind,"
+ " false AS inhdetachpending, NULL\n"
"FROM pg_catalog.pg_class c, pg_catalog.pg_inherits i\n"
"WHERE c.oid = i.inhrelid AND i.inhparent = '%s'\n"
"ORDER BY c.relname;",
@@ -3292,11 +3312,13 @@ describeOneTableDetails(const char *schemaname,
else
printfPQExpBuffer(&buf, "%*s %s",
ctw, "", PQgetvalue(result, i, 0));
- if (!PQgetisnull(result, i, 2))
- appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 2));
+ if (!PQgetisnull(result, i, 3))
+ appendPQExpBuffer(&buf, " %s", PQgetvalue(result, i, 3));
if (child_relkind == RELKIND_PARTITIONED_TABLE ||
child_relkind == RELKIND_PARTITIONED_INDEX)
appendPQExpBufferStr(&buf, ", PARTITIONED");
+ if (strcmp(PQgetvalue(result, i, 2), "t") == 0)
+ appendPQExpBuffer(&buf, " (DETACH PENDING)");
if (i < tuples - 1)
appendPQExpBufferChar(&buf, ',');
diff --git a/src/include/catalog/partition.h b/src/include/catalog/partition.h
index fe3f66befa..c8c7bc1d99 100644
--- a/src/include/catalog/partition.h
+++ b/src/include/catalog/partition.h
@@ -19,7 +19,7 @@
/* Seed for the extended hash function */
#define HASH_PARTITION_SEED UINT64CONST(0x7A5B22367996DCFD)
-extern Oid get_partition_parent(Oid relid);
+extern Oid get_partition_parent(Oid relid, bool even_if_detached);
extern List *get_partition_ancestors(Oid relid);
extern Oid index_get_partition(Relation partition, Oid indexId);
extern List *map_partition_varattnos(List *expr, int fromrel_varno,
diff --git a/src/include/catalog/pg_inherits.h b/src/include/catalog/pg_inherits.h
index 2b71cad9a2..6d07e1b302 100644
--- a/src/include/catalog/pg_inherits.h
+++ b/src/include/catalog/pg_inherits.h
@@ -34,6 +34,7 @@ CATALOG(pg_inherits,2611,InheritsRelationId)
Oid inhrelid BKI_LOOKUP(pg_class);
Oid inhparent BKI_LOOKUP(pg_class);
int32 inhseqno;
+ bool inhdetachpending;
} FormData_pg_inherits;
/* ----------------
@@ -49,7 +50,8 @@ DECLARE_INDEX(pg_inherits_parent_index, 2187, on pg_inherits using btree(inhpare
#define InheritsParentIndexId 2187
-extern List *find_inheritance_children(Oid parentrelId, LOCKMODE lockmode);
+extern List *find_inheritance_children(Oid parentrelId, bool include_detached,
+ LOCKMODE lockmode);
extern List *find_all_inheritors(Oid parentrelId, LOCKMODE lockmode,
List **parents);
extern bool has_subclass(Oid relationId);
@@ -57,6 +59,8 @@ extern bool has_superclass(Oid relationId);
extern bool typeInheritsFrom(Oid subclassTypeId, Oid superclassTypeId);
extern void StoreSingleInheritance(Oid relationId, Oid parentOid,
int32 seqNumber);
-extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent);
+extern bool DeleteInheritsTuple(Oid inhrelid, Oid inhparent, bool allow_detached,
+ const char *childname);
+extern bool PartitionHasPendingDetach(Oid partoid);
#endif /* PG_INHERITS_H */
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index 68425eb2c0..0ce19d98ec 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -871,6 +871,7 @@ typedef struct PartitionCmd
NodeTag type;
RangeVar *name; /* name of partition to attach/detach */
PartitionBoundSpec *bound; /* FOR VALUES, if attaching */
+ bool concurrent;
} PartitionCmd;
/****************************************************************************
@@ -1909,6 +1910,7 @@ typedef enum AlterTableType
AT_GenericOptions, /* OPTIONS (...) */
AT_AttachPartition, /* ATTACH PARTITION */
AT_DetachPartition, /* DETACH PARTITION */
+ AT_DetachPartitionFinalize, /* DETACH PARTITION FINALIZE */
AT_AddIdentity, /* ADD IDENTITY */
AT_SetIdentity, /* SET identity column options */
AT_DropIdentity, /* DROP IDENTITY */
diff --git a/src/include/parser/kwlist.h b/src/include/parser/kwlist.h
index ca1f950cbe..4bbe53e852 100644
--- a/src/include/parser/kwlist.h
+++ b/src/include/parser/kwlist.h
@@ -166,6 +166,7 @@ PG_KEYWORD("false", FALSE_P, RESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("family", FAMILY, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("fetch", FETCH, RESERVED_KEYWORD, AS_LABEL)
PG_KEYWORD("filter", FILTER, UNRESERVED_KEYWORD, AS_LABEL)
+PG_KEYWORD("finalize", FINALIZE, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("first", FIRST_P, UNRESERVED_KEYWORD, BARE_LABEL)
PG_KEYWORD("float", FLOAT_P, COL_NAME_KEYWORD, BARE_LABEL)
PG_KEYWORD("following", FOLLOWING, UNRESERVED_KEYWORD, BARE_LABEL)
diff --git a/src/include/partitioning/partdesc.h b/src/include/partitioning/partdesc.h
index ff113199e5..7f03ff4271 100644
--- a/src/include/partitioning/partdesc.h
+++ b/src/include/partitioning/partdesc.h
@@ -21,6 +21,7 @@
typedef struct PartitionDescData
{
int nparts; /* Number of partitions */
+ bool includes_detached; /* Does it include detached partitions */
Oid *oids; /* Array of 'nparts' elements containing
* partition OIDs in order of the their bounds */
bool *is_leaf; /* Array of 'nparts' elements storing whether
@@ -30,9 +31,9 @@ typedef struct PartitionDescData
} PartitionDescData;
-extern PartitionDesc RelationGetPartitionDesc(Relation rel);
+extern PartitionDesc RelationGetPartitionDesc(Relation rel, bool include_detached);
-extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt);
+extern PartitionDirectory CreatePartitionDirectory(MemoryContext mcxt, bool include_detached);
extern PartitionDesc PartitionDirectoryLookup(PartitionDirectory, Relation);
extern void DestroyPartitionDirectory(PartitionDirectory pdir);
diff --git a/src/include/utils/snapmgr.h b/src/include/utils/snapmgr.h
index f66ac58188..44539fe15a 100644
--- a/src/include/utils/snapmgr.h
+++ b/src/include/utils/snapmgr.h
@@ -132,6 +132,7 @@ extern void AtEOXact_Snapshot(bool isCommit, bool resetXmin);
extern void ImportSnapshot(const char *idstr);
extern bool XactHasExportedSnapshots(void);
extern void DeleteAllExportedSnapshotFiles(void);
+extern void WaitForOlderSnapshots(TransactionId limitXmin, bool progress);
extern bool ThereAreNoPriorRegisteredSnapshots(void);
extern bool TransactionIdLimitedForOldSnapshots(TransactionId recentXmin,
Relation relation,
diff --git a/src/test/isolation/expected/detach-partition-concurrently-1.out b/src/test/isolation/expected/detach-partition-concurrently-1.out
new file mode 100644
index 0000000000..9b4526773e
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-1.out
@@ -0,0 +1,234 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2detach s1s s1c s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1s s3s s3i s1c s3i s2drop s1s
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+f
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s3i: SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2';
+?column?
+
+t
+step s2drop: DROP TABLE d_listp2;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+
+starting permutation: s1b s1s s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1s s1ins2 s2detach s1ins s1s s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1ins s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1ins: INSERT INTO d_listp VALUES (1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1s s2detach s1s s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1b s1ins2 s2detach s3ins2 s1c
+step s1b: BEGIN;
+step s1ins2: INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s3ins2: INSERT INTO d_listp VALUES (2);
+ERROR: no partition of relation "d_listp" found for row
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec1 s3s s1dealloc s1c
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec1: EXECUTE f(1);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1dealloc: DEALLOCATE f;
+step s1c: COMMIT;
+step s2detach: <... completed>
+
+starting permutation: s1brr s1prep s1exec2 s2detach s1s s1exec2 s3s s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1exec2: EXECUTE f(2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+2
+step s1exec2: EXECUTE f(2);
+step s3s: SELECT * FROM d_listp;
+a
+
+1
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s1s s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep: PREPARE f(int) AS INSERT INTO d_listp VALUES ($1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep1 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep1: PREPARE f(int) AS INSERT INTO d_listp VALUES (1);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
+
+starting permutation: s1brr s1prep2 s2detach s1s s1exec2 s1c s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prep2: PREPARE f(int) AS INSERT INTO d_listp VALUES (2);
+step s2detach: ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; <waiting ...>
+step s1s: SELECT * FROM d_listp;
+a
+
+1
+2
+step s1exec2: EXECUTE f(2);
+step s1c: COMMIT;
+step s2detach: <... completed>
+step s1dealloc: DEALLOCATE f;
diff --git a/src/test/isolation/expected/detach-partition-concurrently-2.out b/src/test/isolation/expected/detach-partition-concurrently-2.out
new file mode 100644
index 0000000000..85be707b40
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-2.out
@@ -0,0 +1,66 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s1b s1s s2d s3i1 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s2d s3i2 s3i2 s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3i1 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY;
+ERROR: removing partition "d_lp_fk_1" violates foreign key constraint "d_lp_fk_r_a_fkey1"
+step s1c: COMMIT;
+
+starting permutation: s1b s1s s3i2 s2d s1c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3i2: INSERT INTO d_lp_fk_r VALUES (2);
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s1c: COMMIT;
+step s2d: <... completed>
+
+starting permutation: s1b s1s s3b s2d s3i1 s1c s3c
+step s1b: BEGIN;
+step s1s: SELECT * FROM d_lp_fk;
+a
+
+1
+2
+step s3b: BEGIN;
+step s2d: ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; <waiting ...>
+step s3i1: INSERT INTO d_lp_fk_r VALUES (1);
+ERROR: insert or update on table "d_lp_fk_r" violates foreign key constraint "d_lp_fk_r_a_fkey"
+step s1c: COMMIT;
+step s2d: <... completed>
+step s3c: COMMIT;
diff --git a/src/test/isolation/expected/detach-partition-concurrently-3.out b/src/test/isolation/expected/detach-partition-concurrently-3.out
new file mode 100644
index 0000000000..88e83638c7
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-3.out
@@ -0,0 +1,282 @@
+Parsed test spec with 2 sessions
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1describe s1alter
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1describe: SELECT 'd3_listp' AS root, * FROM pg_partition_tree('d3_listp')
+ UNION ALL SELECT 'd3_listp1', * FROM pg_partition_tree('d3_listp1');
+root relid parentrelid isleaf level
+
+d3_listp d3_listp f 0
+d3_listp1 d3_listp1 t 0
+step s1alter: ALTER TABLE d3_listp1 ALTER a DROP NOT NULL;
+ERROR: cannot alter partition "d3_listp1" with an incomplete detach
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1insert s1c
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1insert: INSERT INTO d3_listp VALUES (1);
+ERROR: no partition of relation "d3_listp" found for row
+step s1c: COMMIT;
+
+starting permutation: s2snitch s1brr s1s s2detach s1cancel s1insert s1c s1spart
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1insert: INSERT INTO d3_listp VALUES (1);
+step s1c: COMMIT;
+step s1spart: SELECT * FROM d3_listp1;
+a
+
+1
+1
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1insertpart
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1insertpart: INSERT INTO d3_listp1 VALUES (1);
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1drop s1list
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1drop: DROP TABLE d3_listp;
+step s1list: SELECT relname FROM pg_catalog.pg_class
+ WHERE relname LIKE 'd3_listp%' ORDER BY 1;
+relname
+
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1trunc s1spart
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1trunc: TRUNCATE TABLE d3_listp;
+step s1spart: SELECT * FROM d3_listp1;
+a
+
+1
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s2begin s2drop s1s s2commit
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s2begin: BEGIN;
+step s2drop: DROP TABLE d3_listp1;
+step s1s: SELECT * FROM d3_listp; <waiting ...>
+step s2commit: COMMIT;
+step s1s: <... completed>
+a
+
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1b s1spart s2detachfinal s1c
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1b: BEGIN;
+step s1spart: SELECT * FROM d3_listp1;
+a
+
+1
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE; <waiting ...>
+step s1c: COMMIT;
+step s2detachfinal: <... completed>
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1b s1s s2detachfinal s1c
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE;
+step s1c: COMMIT;
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s1b s1spart s2detachfinal s1c
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s1b: BEGIN;
+step s1spart: SELECT * FROM d3_listp1;
+a
+
+1
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE; <waiting ...>
+step s1c: COMMIT;
+step s2detachfinal: <... completed>
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s2begin s2detachfinal s2commit
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s2begin: BEGIN;
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE;
+step s2commit: COMMIT;
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s2begin s2detachfinal s1spart s2commit
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s2begin: BEGIN;
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE;
+step s1spart: SELECT * FROM d3_listp1; <waiting ...>
+step s2commit: COMMIT;
+step s1spart: <... completed>
+a
+
+1
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1c s2begin s2detachfinal s1insertpart s2commit
+step s2snitch: INSERT INTO d3_pid SELECT pg_backend_pid();
+step s1b: BEGIN;
+step s1s: SELECT * FROM d3_listp;
+a
+
+1
+step s2detach: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; <waiting ...>
+step s1cancel: SELECT pg_cancel_backend(pid) FROM d3_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1c: COMMIT;
+step s2begin: BEGIN;
+step s2detachfinal: ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE;
+step s1insertpart: INSERT INTO d3_listp1 VALUES (1); <waiting ...>
+step s2commit: COMMIT;
+step s1insertpart: <... completed>
+unused step name: s1droppart
diff --git a/src/test/isolation/expected/detach-partition-concurrently-4.out b/src/test/isolation/expected/detach-partition-concurrently-4.out
new file mode 100644
index 0000000000..e52ed99aa6
--- /dev/null
+++ b/src/test/isolation/expected/detach-partition-concurrently-4.out
@@ -0,0 +1,69 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s2snitch s1b s1s s2detach s1cancel s1insert s1c
+step s2snitch: insert into d4_pid select pg_backend_pid();
+step s1b: begin;
+step s1s: select * from d4_primary;
+a
+
+1
+step s2detach: alter table d4_primary detach partition d4_primary1 concurrently; <waiting ...>
+step s1cancel: select pg_cancel_backend(pid) from d4_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1insert: insert into d4_fk values (1);
+ERROR: insert or update on table "d4_fk" violates foreign key constraint "d4_fk_a_fkey"
+step s1c: commit;
+
+starting permutation: s2snitch s1brr s1s s2detach s1cancel s1insert s1c
+step s2snitch: insert into d4_pid select pg_backend_pid();
+step s1brr: begin isolation level repeatable read;
+step s1s: select * from d4_primary;
+a
+
+1
+step s2detach: alter table d4_primary detach partition d4_primary1 concurrently; <waiting ...>
+step s1cancel: select pg_cancel_backend(pid) from d4_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
+step s1insert: insert into d4_fk values (1);
+ERROR: insert or update on table "d4_fk" violates foreign key constraint "d4_fk_a_fkey"
+step s1c: commit;
+
+starting permutation: s2snitch s1b s1s s2detach s3insert s1c
+step s2snitch: insert into d4_pid select pg_backend_pid();
+step s1b: begin;
+step s1s: select * from d4_primary;
+a
+
+1
+step s2detach: alter table d4_primary detach partition d4_primary1 concurrently; <waiting ...>
+step s3insert: insert into d4_fk values (1);
+ERROR: insert or update on table "d4_fk" violates foreign key constraint "d4_fk_a_fkey"
+step s1c: commit;
+step s2detach: <... completed>
+
+starting permutation: s2snitch s1b s1s s2detach s3brr s3insert s3commit s1cancel
+step s2snitch: insert into d4_pid select pg_backend_pid();
+step s1b: begin;
+step s1s: select * from d4_primary;
+a
+
+1
+step s2detach: alter table d4_primary detach partition d4_primary1 concurrently; <waiting ...>
+step s3brr: begin isolation level repeatable read;
+step s3insert: insert into d4_fk values (1);
+ERROR: insert or update on table "d4_fk" violates foreign key constraint "d4_fk_a_fkey"
+step s3commit: commit;
+step s1cancel: select pg_cancel_backend(pid) from d4_pid;
+pg_cancel_backend
+
+t
+step s2detach: <... completed>
+error in steps s1cancel s2detach: ERROR: canceling statement due to user request
diff --git a/src/test/isolation/isolation_schedule b/src/test/isolation/isolation_schedule
index 5d6b79e66e..f4c01006fc 100644
--- a/src/test/isolation/isolation_schedule
+++ b/src/test/isolation/isolation_schedule
@@ -24,6 +24,10 @@ test: deadlock-hard
test: deadlock-soft
test: deadlock-soft-2
test: deadlock-parallel
+test: detach-partition-concurrently-1
+test: detach-partition-concurrently-2
+test: detach-partition-concurrently-3
+test: detach-partition-concurrently-4
test: fk-contention
test: fk-deadlock
test: fk-deadlock2
diff --git a/src/test/isolation/specs/detach-partition-concurrently-1.spec b/src/test/isolation/specs/detach-partition-concurrently-1.spec
new file mode 100644
index 0000000000..6ca0adac1d
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-1.spec
@@ -0,0 +1,69 @@
+# Test that detach partition concurrently makes the partition invisible at the
+# correct time.
+
+setup
+{
+ DROP TABLE IF EXISTS d_listp, d_listp1, d_listp2;
+ CREATE TABLE d_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d_listp1 PARTITION OF d_listp FOR VALUES IN (1);
+ CREATE TABLE d_listp2 PARTITION OF d_listp FOR VALUES IN (2);
+ INSERT INTO d_listp VALUES (1),(2);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d_listp, d_listp2, d_listp_foobar;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d_listp; }
+step "s1ins" { INSERT INTO d_listp VALUES (1); }
+step "s1ins2" { INSERT INTO d_listp VALUES (2); }
+step "s1prep" { PREPARE f(int) AS INSERT INTO d_listp VALUES ($1); }
+step "s1prep1" { PREPARE f(int) AS INSERT INTO d_listp VALUES (1); }
+step "s1prep2" { PREPARE f(int) AS INSERT INTO d_listp VALUES (2); }
+step "s1exec1" { EXECUTE f(1); }
+step "s1exec2" { EXECUTE f(2); }
+step "s1dealloc" { DEALLOCATE f; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2detach" { ALTER TABLE d_listp DETACH PARTITION d_listp2 CONCURRENTLY; }
+step "s2drop" { DROP TABLE d_listp2; }
+
+session "s3"
+step "s3s" { SELECT * FROM d_listp; }
+step "s3i" { SELECT relpartbound IS NULL FROM pg_class where relname = 'd_listp2'; }
+step "s3ins2" { INSERT INTO d_listp VALUES (2); }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate, as does anybody else.
+permutation "s1b" "s1s" "s2detach" "s1s" "s1c" "s1s"
+
+# relpartbound remains set until s1 commits
+# XXX this could be timing dependent :-(
+permutation "s1b" "s1s" "s2detach" "s1s" "s3s" "s3i" "s1c" "s3i" "s2drop" "s1s"
+
+# In read-committed mode, the partition disappears from view of concurrent
+# transactions immediately. But if a write lock is held, then the detach
+# has to wait.
+permutation "s1b" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1b" "s1s" "s1ins2" "s2detach" "s1ins" "s1s" "s1c"
+
+# In repeatable-read mode, the partition remains visible until commit even
+# if the to-be-detached partition is not locked for write.
+permutation "s1brr" "s1s" "s2detach" "s1ins" "s1s" "s1c"
+permutation "s1brr" "s1s" "s2detach" "s1s" "s1c"
+
+# Another process trying to acquire a write lock will be blocked behind the
+# detacher
+permutation "s1b" "s1ins2" "s2detach" "s3ins2" "s1c"
+
+# a prepared query is not blocked
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec1" "s3s" "s1dealloc" "s1c"
+permutation "s1brr" "s1prep" "s1exec2" "s2detach" "s1s" "s1exec2" "s3s" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s1s" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep1" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
+permutation "s1brr" "s1prep2" "s2detach" "s1s" "s1exec2" "s1c" "s1dealloc"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-2.spec b/src/test/isolation/specs/detach-partition-concurrently-2.spec
new file mode 100644
index 0000000000..9281c80a69
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-2.spec
@@ -0,0 +1,41 @@
+# Test that detach partition concurrently makes the partition safe
+# for foreign keys that reference it.
+
+setup
+{
+ DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r;
+
+ CREATE TABLE d_lp_fk (a int PRIMARY KEY) PARTITION BY LIST(a);
+ CREATE TABLE d_lp_fk_1 PARTITION OF d_lp_fk FOR VALUES IN (1);
+ CREATE TABLE d_lp_fk_2 PARTITION OF d_lp_fk FOR VALUES IN (2);
+ INSERT INTO d_lp_fk VALUES (1), (2);
+
+ CREATE TABLE d_lp_fk_r (a int references d_lp_fk);
+}
+
+teardown { DROP TABLE IF EXISTS d_lp_fk, d_lp_fk_1, d_lp_fk_2, d_lp_fk_r; }
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1s" { SELECT * FROM d_lp_fk; }
+step "s1c" { COMMIT; }
+
+session "s2"
+step "s2d" { ALTER TABLE d_lp_fk DETACH PARTITION d_lp_fk_1 CONCURRENTLY; }
+
+session "s3"
+step "s3b" { BEGIN; }
+step "s3i1" { INSERT INTO d_lp_fk_r VALUES (1); }
+step "s3i2" { INSERT INTO d_lp_fk_r VALUES (2); }
+step "s3c" { COMMIT; }
+
+# The transaction that detaches hangs until it sees any older transaction
+# terminate.
+permutation "s1b" "s1s" "s2d" "s3i1" "s1c"
+permutation "s1b" "s1s" "s2d" "s3i2" "s3i2" "s1c"
+
+permutation "s1b" "s1s" "s3i1" "s2d" "s1c"
+permutation "s1b" "s1s" "s3i2" "s2d" "s1c"
+
+# what if s3 has an uncommitted insertion?
+permutation "s1b" "s1s" "s3b" "s2d" "s3i1" "s1c" "s3c"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-3.spec b/src/test/isolation/specs/detach-partition-concurrently-3.spec
new file mode 100644
index 0000000000..4b706430e1
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-3.spec
@@ -0,0 +1,66 @@
+# Try various things to happen to a partition with an incomplete detach
+
+setup
+{
+ CREATE TABLE d3_listp (a int) PARTITION BY LIST(a);
+ CREATE TABLE d3_listp1 PARTITION OF d3_listp FOR VALUES IN (1);
+ CREATE TABLE d3_pid (pid int);
+ INSERT INTO d3_listp VALUES (1);
+}
+
+teardown {
+ DROP TABLE IF EXISTS d3_listp, d3_listp1, d3_pid;
+}
+
+session "s1"
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1s" { SELECT * FROM d3_listp; }
+step "s1spart" { SELECT * FROM d3_listp1; }
+step "s1cancel" { SELECT pg_cancel_backend(pid) FROM d3_pid; }
+step "s1c" { COMMIT; }
+step "s1alter" { ALTER TABLE d3_listp1 ALTER a DROP NOT NULL; }
+step "s1insert" { INSERT INTO d3_listp VALUES (1); }
+step "s1insertpart" { INSERT INTO d3_listp1 VALUES (1); }
+step "s1drop" { DROP TABLE d3_listp; }
+step "s1droppart" { DROP TABLE d3_listp1; }
+step "s1trunc" { TRUNCATE TABLE d3_listp; }
+step "s1list" { SELECT relname FROM pg_catalog.pg_class
+ WHERE relname LIKE 'd3_listp%' ORDER BY 1; }
+step "s1describe" { SELECT 'd3_listp' AS root, * FROM pg_partition_tree('d3_listp')
+ UNION ALL SELECT 'd3_listp1', * FROM pg_partition_tree('d3_listp1'); }
+
+session "s2"
+step "s2begin" { BEGIN; }
+step "s2snitch" { INSERT INTO d3_pid SELECT pg_backend_pid(); }
+step "s2detach" { ALTER TABLE d3_listp DETACH PARTITION d3_listp1 CONCURRENTLY; }
+step "s2detachfinal" { ALTER TABLE d3_listp DETACH PARTITION d3_listp1 FINALIZE; }
+step "s2drop" { DROP TABLE d3_listp1; }
+step "s2commit" { COMMIT; }
+
+# Try various things while the partition is in "being detached" state, with
+# no session waiting.
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1describe" "s1alter"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1insert" "s1c"
+permutation "s2snitch" "s1brr" "s1s" "s2detach" "s1cancel" "s1insert" "s1c" "s1spart"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1insertpart"
+# "drop" here does both tables
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1drop" "s1list"
+# "truncate" only does parent, not partition
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1trunc" "s1spart"
+
+# When a partition with incomplete detach is dropped, we grab lock on parent too.
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s2begin" "s2drop" "s1s" "s2commit"
+
+# Partially detach, then select and try to complete the detach. Reading
+# from partition blocks (AEL is required on partition); reading from parent
+# does not block.
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1b" "s1spart" "s2detachfinal" "s1c"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1b" "s1s" "s2detachfinal" "s1c"
+
+# DETACH FINALIZE in a transaction block. No insert/select on the partition
+# is allowed concurrently with that.
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s1b" "s1spart" "s2detachfinal" "s1c"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s2begin" "s2detachfinal" "s2commit"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s2begin" "s2detachfinal" "s1spart" "s2commit"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1c" "s2begin" "s2detachfinal" "s1insertpart" "s2commit"
diff --git a/src/test/isolation/specs/detach-partition-concurrently-4.spec b/src/test/isolation/specs/detach-partition-concurrently-4.spec
new file mode 100644
index 0000000000..fa72386da7
--- /dev/null
+++ b/src/test/isolation/specs/detach-partition-concurrently-4.spec
@@ -0,0 +1,36 @@
+# This test exercises behavior of foreign keys in the face of concurrent
+# detach of partitions in the referenced table.
+setup {
+ drop table if exists d4_primary, d4_primary1, d4_fk, d4_pid;
+ create table d4_primary (a int primary key) partition by list (a);
+ create table d4_primary1 partition of d4_primary for values in (1);
+ insert into d4_primary values (1);
+ create table d4_fk (a int references d4_primary);
+ create table d4_pid (pid int);
+}
+
+session "s1"
+step "s1b" { begin; }
+step "s1brr" { begin isolation level repeatable read; }
+step "s1s" { select * from d4_primary; }
+step "s1cancel" { select pg_cancel_backend(pid) from d4_pid; }
+step "s1insert" { insert into d4_fk values (1); }
+step "s1c" { commit; }
+
+session "s2"
+step "s2snitch" { insert into d4_pid select pg_backend_pid(); }
+step "s2detach" { alter table d4_primary detach partition d4_primary1 concurrently; }
+
+session "s3"
+step "s3brr" { begin isolation level repeatable read; }
+step "s3insert" { insert into d4_fk values (1); }
+step "s3commit" { commit; }
+
+# Trying to insert into a partially detached partition is rejected
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s1cancel" "s1insert" "s1c"
+# ... even under REPEATABLE READ mode.
+permutation "s2snitch" "s1brr" "s1s" "s2detach" "s1cancel" "s1insert" "s1c"
+
+# These other cases are fairly trivial.
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s3insert" "s1c"
+permutation "s2snitch" "s1b" "s1s" "s2detach" "s3brr" "s3insert" "s3commit" "s1cancel"
diff --git a/src/test/modules/delay_execution/Makefile b/src/test/modules/delay_execution/Makefile
index f270aebf3a..70f24e846d 100644
--- a/src/test/modules/delay_execution/Makefile
+++ b/src/test/modules/delay_execution/Makefile
@@ -7,7 +7,8 @@ OBJS = \
$(WIN32RES) \
delay_execution.o
-ISOLATION = partition-addition
+ISOLATION = partition-addition \
+ partition-removal-1
ifdef USE_PGXS
PG_CONFIG = pg_config
diff --git a/src/test/modules/delay_execution/expected/partition-removal-1.out b/src/test/modules/delay_execution/expected/partition-removal-1.out
new file mode 100644
index 0000000000..427f41c9aa
--- /dev/null
+++ b/src/test/modules/delay_execution/expected/partition-removal-1.out
@@ -0,0 +1,175 @@
+Parsed test spec with 3 sessions
+
+starting permutation: s3lock s1b s1exec s2remp s3check s3unlock s3check s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3check: SELECT * FROM partrem;
+a b
+
+1 ABC
+3 DEF
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s3check: SELECT * FROM partrem;
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec s2remp s3check s3unlock s3check s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec: SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3check: SELECT * FROM partrem;
+a b
+
+1 ABC
+3 DEF
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec: <... completed>
+a b
+
+2 JKL
+step s3check: SELECT * FROM partrem;
+a b
+
+1 ABC
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1b s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1b: BEGIN;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2) AND a <> 1; <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1exec2 s2remp s3unlock s1c
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1exec2: SELECT * FROM partrem WHERE a <> (SELECT 2) AND a <> 1; <waiting ...>
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1exec2: <... completed>
+a b
+
+3 DEF
+step s1c: COMMIT;
+step s2remp: <... completed>
+
+starting permutation: s3lock s1brr s1prepare s2remp s1execprep s3unlock s1check s1c s1check s1dealloc
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prepare: PREPARE ins AS INSERT INTO partrem VALUES ($1, 'GHI');
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s1execprep: EXECUTE ins(2); <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1execprep: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+2 GHI
+step s1c: COMMIT;
+step s2remp: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+step s1dealloc: DEALLOCATE ins;
+
+starting permutation: s1brr s1prepare s2remp s3lock s1execprep s3unlock s1check s1c s1check s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1prepare: PREPARE ins AS INSERT INTO partrem VALUES ($1, 'GHI');
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s1execprep: EXECUTE ins(2); <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1execprep: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+2 GHI
+step s1c: COMMIT;
+step s2remp: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+step s1dealloc: DEALLOCATE ins;
+
+starting permutation: s1brr s1check s3lock s2remp s1prepare s1execprep s3unlock s1check s1c s1check s1dealloc
+step s1brr: BEGIN ISOLATION LEVEL REPEATABLE READ;
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+step s3lock: SELECT pg_advisory_lock(12543);
+pg_advisory_lock
+
+
+step s2remp: ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; <waiting ...>
+step s1prepare: PREPARE ins AS INSERT INTO partrem VALUES ($1, 'GHI');
+step s1execprep: EXECUTE ins(2); <waiting ...>
+step s3unlock: SELECT pg_advisory_unlock(12543);
+pg_advisory_unlock
+
+t
+step s1execprep: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+2 GHI
+step s1c: COMMIT;
+step s2remp: <... completed>
+step s1check: SELECT * FROM partrem WHERE b = 'GHI';
+a b
+
+step s1dealloc: DEALLOCATE ins;
diff --git a/src/test/modules/delay_execution/specs/partition-removal-1.spec b/src/test/modules/delay_execution/specs/partition-removal-1.spec
new file mode 100644
index 0000000000..5ee2750129
--- /dev/null
+++ b/src/test/modules/delay_execution/specs/partition-removal-1.spec
@@ -0,0 +1,58 @@
+# Test removal of a partition with less-than-exclusive locking.
+
+setup
+{
+ CREATE TABLE partrem (a int, b text) PARTITION BY LIST(a);
+ CREATE TABLE partrem1 PARTITION OF partrem FOR VALUES IN (1);
+ CREATE TABLE partrem2 PARTITION OF partrem FOR VALUES IN (2);
+ CREATE TABLE partrem3 PARTITION OF partrem FOR VALUES IN (3);
+ INSERT INTO partrem VALUES (1, 'ABC');
+ INSERT INTO partrem VALUES (2, 'JKL');
+ INSERT INTO partrem VALUES (3, 'DEF');
+}
+
+teardown
+{
+ DROP TABLE IF EXISTS partrem, partrem2;
+}
+
+session "s1"
+setup { LOAD 'delay_execution';
+ SET delay_execution.post_planning_lock_id = 12543; }
+step "s1b" { BEGIN; }
+step "s1brr" { BEGIN ISOLATION LEVEL REPEATABLE READ; }
+step "s1exec" { SELECT * FROM partrem WHERE a <> 1 AND a <> (SELECT 3); }
+step "s1exec2" { SELECT * FROM partrem WHERE a <> (SELECT 2) AND a <> 1; }
+step "s1prepare" { PREPARE ins AS INSERT INTO partrem VALUES ($1, 'GHI'); }
+step "s1execprep" { EXECUTE ins(2); }
+step "s1check" { SELECT * FROM partrem WHERE b = 'GHI'; }
+step "s1c" { COMMIT; }
+step "s1dealloc" { DEALLOCATE ins; }
+
+session "s2"
+step "s2remp" { ALTER TABLE partrem DETACH PARTITION partrem2 CONCURRENTLY; }
+
+session "s3"
+step "s3lock" { SELECT pg_advisory_lock(12543); }
+step "s3unlock" { SELECT pg_advisory_unlock(12543); }
+step "s3check" { SELECT * FROM partrem; }
+
+# The SELECT will be planned with all three partitions shown above,
+# of which we expect partrem1 to be pruned at planning and partrem3 at
+# execution. Then we'll block, and by the time the query is actually
+# executed, detach of partrem2 is already underway (so its row doesn't
+# show up in s3's result), but we expect its row to still appear in the
+# result for s1.
+permutation "s3lock" "s1b" "s1exec" "s2remp" "s3check" "s3unlock" "s3check" "s1c"
+permutation "s3lock" "s1brr" "s1exec" "s2remp" "s3check" "s3unlock" "s3check" "s1c"
+
+# In this case we're testing that after pruning partrem2 at runtime, the
+# query still works correctly.
+permutation "s3lock" "s1b" "s1exec2" "s2remp" "s3unlock" "s1c"
+permutation "s3lock" "s1brr" "s1exec2" "s2remp" "s3unlock" "s1c"
+
+# In this case we test that an insert that's prepared in repeatable read
+# mode still works after detaching.
+permutation "s3lock" "s1brr" "s1prepare" "s2remp" "s1execprep" "s3unlock" "s1check" "s1c" "s1check" "s1dealloc"
+permutation "s1brr" "s1prepare" "s2remp" "s3lock" "s1execprep" "s3unlock" "s1check" "s1c" "s1check" "s1dealloc"
+permutation "s1brr" "s1check" "s3lock" "s2remp" "s1prepare" "s1execprep" "s3unlock" "s1check" "s1c" "s1check" "s1dealloc"
diff --git a/src/test/regress/expected/alter_table.out b/src/test/regress/expected/alter_table.out
index bb3f873f22..ec14b060a6 100644
--- a/src/test/regress/expected/alter_table.out
+++ b/src/test/regress/expected/alter_table.out
@@ -4163,6 +4163,35 @@ SELECT * from part_rp;
(0 rows)
DROP TABLE part_rp;
+-- concurrent detach
+CREATE TABLE range_parted2 (
+ a int
+) PARTITION BY RANGE(a);
+CREATE TABLE part_rp PARTITION OF range_parted2 FOR VALUES FROM (0) to (100);
+BEGIN;
+-- doesn't work in a partition block
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+ERROR: ALTER TABLE ... DETACH CONCURRENTLY cannot run inside a transaction block
+COMMIT;
+CREATE TABLE part_rpd PARTITION OF range_parted2 DEFAULT;
+-- doesn't work if there's a default partition
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+ERROR: cannot detach partitions concurrently when a default partition exists
+-- doesn't work for the default partition
+ALTER TABLE range_parted2 DETACH PARTITION part_rpd CONCURRENTLY;
+ERROR: cannot detach partitions concurrently when a default partition exists
+DROP TABLE part_rpd;
+-- works fine
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+\d+ range_parted2
+ Partitioned table "public.range_parted2"
+ Column | Type | Collation | Nullable | Default | Storage | Stats target | Description
+--------+---------+-----------+----------+---------+---------+--------------+-------------
+ a | integer | | | | plain | |
+Partition key: RANGE (a)
+Number of partitions: 0
+
+DROP TABLE range_parted2;
-- Check ALTER TABLE commands for partitioned tables and partitions
-- cannot add/drop column to/from *only* the parent
ALTER TABLE ONLY list_parted2 ADD COLUMN c int;
diff --git a/src/test/regress/sql/alter_table.sql b/src/test/regress/sql/alter_table.sql
index 4cc55d8525..7a9c9252dc 100644
--- a/src/test/regress/sql/alter_table.sql
+++ b/src/test/regress/sql/alter_table.sql
@@ -2678,6 +2678,26 @@ DROP TABLE range_parted2;
SELECT * from part_rp;
DROP TABLE part_rp;
+-- concurrent detach
+CREATE TABLE range_parted2 (
+ a int
+) PARTITION BY RANGE(a);
+CREATE TABLE part_rp PARTITION OF range_parted2 FOR VALUES FROM (0) to (100);
+BEGIN;
+-- doesn't work in a partition block
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+COMMIT;
+CREATE TABLE part_rpd PARTITION OF range_parted2 DEFAULT;
+-- doesn't work if there's a default partition
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+-- doesn't work for the default partition
+ALTER TABLE range_parted2 DETACH PARTITION part_rpd CONCURRENTLY;
+DROP TABLE part_rpd;
+-- works fine
+ALTER TABLE range_parted2 DETACH PARTITION part_rp CONCURRENTLY;
+\d+ range_parted2
+DROP TABLE range_parted2;
+
-- Check ALTER TABLE commands for partitioned tables and partitions
-- cannot add/drop column to/from *only* the parent
--
2.20.1
--IS0zKkzwUGydFO0o--
^ permalink raw reply [nested|flat] 57+ messages in thread
* Re: encoding affects ICU regex character classification
@ 2023-11-30 00:23 Jeff Davis <[email protected]>
2023-11-30 02:10 ` Re: encoding affects ICU regex character classification Thomas Munro <[email protected]>
0 siblings, 1 reply; 57+ messages in thread
From: Jeff Davis @ 2023-11-30 00:23 UTC (permalink / raw)
To: Tom Lane <[email protected]>; +Cc: pgsql-hackers
On Wed, 2023-11-29 at 18:56 -0500, Tom Lane wrote:
> We'd have to find an alternate source of knowledge to replace the
> <wctype.h> functions if we wanted to fix it fully ... can ICU do
> that?
My follow-up proposal is exactly along those lines, except that we
don't even need ICU.
By adding a couple lookup tables generated from the Unicode data files,
we can offer a pg_u_isalpha() family of functions. As a bonus, I have
some exhaustive tests to compare with what ICU does so we can protect
ourselves from simple mistakes.
I might as well send it now; patch attached (0003 is the interesting
one).
I also tested against the iswalpha() family of functions, and those
have very similar behavior (apart from the "C" locale, of course).
Character classification is not localized at all in libc or ICU as far
as I can tell.
There are some differences, and I don't understand why those
differences exist, so perhaps that's worth discussing. Some differences
seem to be related to the titlecase/uppercase distinction. Others are
strange, like how glibc counts some digit characters (outside 0-9) as
alphabetic. And some seem arbitrary, like excluding a few whitespace
characters. I can try to post more details if that would be helpful.
Another issue is that right now we are doing the wrong thing with ICU:
we should be using the u_isUAlphabetic() family of functions, not the
u_isalpha() family of functions.
Regards,
Jeff Davis
Attachments:
[text/x-patch] v1-0003-Add-Unicode-property-tables.patch (85.8K, ../../[email protected]/2-v1-0003-Add-Unicode-property-tables.patch)
download | inline diff:
From f0c004846542f6b415005f9f9c949199b3f3bcfd Mon Sep 17 00:00:00 2001
From: Jeff Davis <[email protected]>
Date: Sat, 18 Nov 2023 15:34:24 -0800
Subject: [PATCH v1 3/3] Add Unicode property tables.
---
src/common/unicode/Makefile | 6 +-
src/common/unicode/category_test.c | 167 +-
.../generate-unicode_category_table.pl | 203 +-
src/common/unicode/meson.build | 4 +-
src/common/unicode_category.c | 210 +-
src/include/common/unicode_category.h | 19 +-
src/include/common/unicode_category_table.h | 2532 +++++++++++++++++
7 files changed, 3078 insertions(+), 63 deletions(-)
diff --git a/src/common/unicode/Makefile b/src/common/unicode/Makefile
index 04d81dd5cb..27f0408d8b 100644
--- a/src/common/unicode/Makefile
+++ b/src/common/unicode/Makefile
@@ -29,13 +29,13 @@ update-unicode: unicode_category_table.h unicode_east_asian_fw_table.h unicode_n
# These files are part of the Unicode Character Database. Download
# them on demand. The dependency on Makefile.global is for
# UNICODE_VERSION.
-CompositionExclusions.txt DerivedNormalizationProps.txt EastAsianWidth.txt NormalizationTest.txt UnicodeData.txt: $(top_builddir)/src/Makefile.global
+CompositionExclusions.txt DerivedCoreProperties.txt DerivedNormalizationProps.txt EastAsianWidth.txt NormalizationTest.txt PropList.txt UnicodeData.txt: $(top_builddir)/src/Makefile.global
$(DOWNLOAD) https://www.unicode.org/Public/$(UNICODE_VERSION)/ucd/$(@F)
unicode_version.h: generate-unicode_version.pl
$(PERL) $< --version $(UNICODE_VERSION)
-unicode_category_table.h: generate-unicode_category_table.pl UnicodeData.txt
+unicode_category_table.h: generate-unicode_category_table.pl DerivedCoreProperties.txt PropList.txt UnicodeData.txt
$(PERL) $<
# Generation of conversion tables used for string normalization with
@@ -82,4 +82,4 @@ clean:
rm -f $(OBJS) category_test category_test.o norm_test norm_test.o
distclean: clean
- rm -f CompositionExclusions.txt DerivedNormalizationProps.txt EastAsianWidth.txt NormalizationTest.txt UnicodeData.txt norm_test_table.h unicode_category_table.h unicode_norm_table.h
+ rm -f CompositionExclusions.txt DerivedCoreProperties.txt DerivedNormalizationProps.txt EastAsianWidth.txt NormalizationTest.txt PropList.txt UnicodeData.txt norm_test_table.h unicode_category_table.h unicode_norm_table.h
diff --git a/src/common/unicode/category_test.c b/src/common/unicode/category_test.c
index d9ea806eb8..02b4f0698a 100644
--- a/src/common/unicode/category_test.c
+++ b/src/common/unicode/category_test.c
@@ -1,6 +1,7 @@
/*-------------------------------------------------------------------------
* category_test.c
- * Program to test Unicode general category functions.
+ * Program to test Unicode general category and character class
+ * functions.
*
* Portions Copyright (c) 2017-2023, PostgreSQL Global Development Group
*
@@ -14,17 +15,21 @@
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
-
#ifdef USE_ICU
#include <unicode/uchar.h>
#endif
-#include "common/unicode_category.h"
+
#include "common/unicode_version.h"
+#include "common/unicode_category.h"
+
+static int pg_unicode_version = 0;
+#ifdef USE_ICU
+static int icu_unicode_version = 0;
+#endif
/*
* Parse version into integer for easy comparison.
*/
-#ifdef USE_ICU
static int
parse_unicode_version(const char *version)
{
@@ -39,56 +44,116 @@ parse_unicode_version(const char *version)
return major * 100 + minor;
}
-#endif
+#ifdef USE_ICU
/*
- * Exhaustively test that the Unicode category for each codepoint matches that
- * returned by ICU.
+ * Test Postgres Unicode tables by comparing with ICU. Test the General
+ * Category, as well as the properties Alphabetic, Lowercase, Uppercase,
+ * White_Space, and Hex_Digit.
*/
-int
-main(int argc, char **argv)
+static void
+icu_test()
{
-#ifdef USE_ICU
- int pg_unicode_version = parse_unicode_version(PG_UNICODE_VERSION);
- int icu_unicode_version = parse_unicode_version(U_UNICODE_VERSION);
int pg_skipped_codepoints = 0;
int icu_skipped_codepoints = 0;
- printf("category_test: Postgres Unicode version:\t%s\n", PG_UNICODE_VERSION);
- printf("category_test: ICU Unicode version:\t\t%s\n", U_UNICODE_VERSION);
-
- for (UChar32 code = 0; code <= 0x10ffff; code++)
+ for (pg_wchar code = 0; code <= 0x10ffff; code++)
{
uint8_t pg_category = unicode_category(code);
uint8_t icu_category = u_charType(code);
+ bool isalpha = pg_u_isalpha(code);
+ bool islower = pg_u_islower(code);
+ bool isupper = pg_u_isupper(code);
+ bool ispunct = pg_u_ispunct(code);
+ bool isdigit = pg_u_isdigit(code);
+ bool isxdigit = pg_u_isxdigit(code);
+ bool isalnum = pg_u_isalnum(code);
+ bool isspace = pg_u_isspace(code);
+ bool isblank = pg_u_isblank(code);
+ bool iscntrl = pg_u_iscntrl(code);
+ bool isgraph = pg_u_isgraph(code);
+ bool isprint = pg_u_isprint(code);
+
+ bool icu_isalpha = u_isUAlphabetic(code);
+ bool icu_islower = u_isULowercase(code);
+ bool icu_isupper = u_isUUppercase(code);
+ bool icu_ispunct = u_ispunct(code);
+ bool icu_isdigit = u_isdigit(code);
+
+ /*
+ * ICU documents that UCHAR_POSIX_XDIGIT should match the xdigit class
+ * in: http://www.unicode.org/reports/tr18/#Compatibility_Properties,
+ * but it does not, so we use UCHAR_HEX_DIGIT instead.
+ */
+ bool icu_isxdigit = u_hasBinaryProperty(code,
+ UCHAR_HEX_DIGIT);
+
+ bool icu_isalnum = u_hasBinaryProperty(code,
+ UCHAR_POSIX_ALNUM);
+ bool icu_isspace = u_isUWhiteSpace(code);
+ bool icu_isblank = u_isblank(code);
+ bool icu_iscntrl = icu_category == PG_U_CONTROL;
+ bool icu_isgraph = u_hasBinaryProperty(code,
+ UCHAR_POSIX_GRAPH);
+ bool icu_isprint = u_hasBinaryProperty(code,
+ UCHAR_POSIX_PRINT);
+
+ /*
+ * A version mismatch means that some assigned codepoints in the newer
+ * version may be unassigned in the older version. That's OK, though
+ * the test will not cover those codepoints marked unassigned in the
+ * older version (that is, it will no longer be an exhaustive test).
+ */
+ if (pg_category == PG_U_UNASSIGNED &&
+ icu_category != PG_U_UNASSIGNED &&
+ pg_unicode_version < icu_unicode_version)
+ {
+ pg_skipped_codepoints++;
+ continue;
+ }
+
+ if (icu_category == PG_U_UNASSIGNED &&
+ pg_category != PG_U_UNASSIGNED &&
+ icu_unicode_version < pg_unicode_version)
+ {
+ icu_skipped_codepoints++;
+ continue;
+ }
+
if (pg_category != icu_category)
{
- /*
- * A version mismatch means that some assigned codepoints in the
- * newer version may be unassigned in the older version. That's
- * OK, though the test will not cover those codepoints marked
- * unassigned in the older version (that is, it will no longer be
- * an exhaustive test).
- */
- if (pg_category == PG_U_UNASSIGNED &&
- pg_unicode_version < icu_unicode_version)
- pg_skipped_codepoints++;
- else if (icu_category == PG_U_UNASSIGNED &&
- icu_unicode_version < pg_unicode_version)
- icu_skipped_codepoints++;
- else
- {
- printf("category_test: FAILURE for codepoint 0x%06x\n", code);
- printf("category_test: Postgres category: %02d %s %s\n", pg_category,
- unicode_category_abbrev(pg_category),
- unicode_category_string(pg_category));
- printf("category_test: ICU category: %02d %s %s\n", icu_category,
- unicode_category_abbrev(icu_category),
- unicode_category_string(icu_category));
- printf("\n");
- exit(1);
- }
+ printf("category_test: FAILURE for codepoint 0x%06x\n", code);
+ printf("category_test: Postgres category: %02d %s %s\n", pg_category,
+ unicode_category_abbrev(pg_category),
+ unicode_category_string(pg_category));
+ printf("category_test: ICU category: %02d %s %s\n", icu_category,
+ unicode_category_abbrev(icu_category),
+ unicode_category_string(icu_category));
+ printf("\n");
+ exit(1);
+ }
+
+ if (isalpha != icu_isalpha ||
+ islower != icu_islower ||
+ isupper != icu_isupper ||
+ ispunct != icu_ispunct ||
+ isdigit != icu_isdigit ||
+ isxdigit != icu_isxdigit ||
+ isalnum != icu_isalnum ||
+ isspace != icu_isspace ||
+ isblank != icu_isblank ||
+ iscntrl != icu_iscntrl ||
+ isgraph != icu_isgraph ||
+ isprint != icu_isprint)
+ {
+ printf("category_test: FAILURE for codepoint 0x%06x\n", code);
+ printf("category_test: Postgres property alpha/lower/upper/punct/digit/xdigit/alnum/space/blank/cntrl/graph/print: %d/%d/%d/%d/%d/%d/%d/%d/%d/%d/%d/%d\n",
+ isalpha, islower, isupper, ispunct, isdigit, isxdigit, isalnum, isspace, isblank, iscntrl, isgraph, isprint);
+ printf("category_test: ICU property alpha/lower/upper/punct/digit/xdigit/alnum/space/blank/cntrl/graph/print: %d/%d/%d/%d/%d/%d/%d/%d/%d/%d/%d/%d\n",
+ icu_isalpha, icu_islower, icu_isupper, icu_ispunct, icu_isdigit, icu_isxdigit, icu_isalnum, icu_isspace, icu_isblank, icu_iscntrl, icu_isgraph, icu_isprint);
+ printf("\n");
+ exit(1);
}
}
@@ -99,10 +164,22 @@ main(int argc, char **argv)
printf("category_test: skipped %d codepoints unassigned in ICU due to Unicode version mismatch\n",
icu_skipped_codepoints);
- printf("category_test: success\n");
- exit(0);
+ printf("category_test: ICU test successful\n");
+}
+#endif
+
+int
+main(int argc, char **argv)
+{
+ pg_unicode_version = parse_unicode_version(PG_UNICODE_VERSION);
+ printf("category_test: Postgres Unicode version:\t%s\n", PG_UNICODE_VERSION);
+
+#ifdef USE_ICU
+ icu_unicode_version = parse_unicode_version(U_UNICODE_VERSION);
+ printf("category_test: ICU Unicode version:\t\t%s\n", U_UNICODE_VERSION);
+
+ icu_test();
#else
- printf("category_test: ICU support required for test; skipping\n");
- exit(0);
+ printf("category_test: ICU not available; skipping\n");
#endif
}
diff --git a/src/common/unicode/generate-unicode_category_table.pl b/src/common/unicode/generate-unicode_category_table.pl
index 992b877ede..9545728443 100644
--- a/src/common/unicode/generate-unicode_category_table.pl
+++ b/src/common/unicode/generate-unicode_category_table.pl
@@ -120,8 +120,6 @@ if ($range_category ne $CATEGORY_UNASSIGNED) {
category => $range_category});
}
-my $num_ranges = scalar @category_ranges;
-
# See: https://www.unicode.org/reports/tr44/#General_Category_Values
my $categories = {
Cn => 'PG_U_UNASSIGNED',
@@ -156,11 +154,98 @@ my $categories = {
Pf => 'PG_U_FINAL_PUNCTUATION'
};
-# Start writing out the output files
+# Find White_Space and Hex_Digit characters
+my @white_space = ();
+my @hex_digits = ();
+my @join_control = ();
+open($FH, '<', "$output_path/PropList.txt")
+ or die "Could not open $output_path/PropList.txt: $!.";
+while (my $line = <$FH>)
+{
+ my $pattern = qr/([0-9A-F\.]+)\s*;\s*(\w+)\s*#.*/s;
+ next unless $line =~ $pattern;
+
+ my $code = $line =~ s/$pattern/$1/rg;
+ my $property = $line =~ s/$pattern/$2/rg;
+ my $start;
+ my $end;
+
+ if ($code =~ /\.\./) {
+ # code range
+ my @sp = split /\.\./, $code;
+ $start = hex($sp[0]);
+ $end = hex($sp[1]);
+ } else {
+ # single code point
+ $start = hex($code);
+ $end = hex($code);
+ }
+
+ if ($property eq "White_Space") {
+ push @white_space, {start => $start, end => $end};
+ }
+ elsif ($property eq "Hex_Digit") {
+ push @hex_digits, {start => $start, end => $end};
+ }
+ elsif ($property eq "Join_Control") {
+ push @join_control, {start => $start, end => $end};
+ }
+}
+
+# Find Alphabetic, Lowercase, and Uppercase characters
+my @alphabetic = ();
+my @lowercase = ();
+my @uppercase = ();
+open($FH, '<', "$output_path/DerivedCoreProperties.txt")
+ or die "Could not open $output_path/DerivedCoreProperties.txt: $!.";
+while (my $line = <$FH>)
+{
+ my $pattern = qr/^([0-9A-F\.]+)\s*;\s*(\w+)\s*#.*$/s;
+ next unless $line =~ $pattern;
+
+ my $code = $line =~ s/$pattern/$1/rg;
+ my $property = $line =~ s/$pattern/$2/rg;
+ my $start;
+ my $end;
+
+ if ($code =~ /\.\./) {
+ # code range
+ my @sp = split /\.\./, $code;
+ die "line: {$line} code: {$code} sp[0] {$sp[0]} sp[1] {$sp[1]}"
+ unless $sp[0] =~ /^[0-9A-F]+$/ && $sp[1] =~ /^[0-9A-F]+$/;
+ $start = hex($sp[0]);
+ $end = hex($sp[1]);
+ } else {
+ die "line: {$line} code: {$code}" unless $code =~ /^[0-9A-F]+$/;
+ # single code point
+ $start = hex($code);
+ $end = hex($code);
+ }
+
+ if ($property eq "Alphabetic") {
+ push @alphabetic, {start => $start, end => $end};
+ }
+ elsif ($property eq "Lowercase") {
+ push @lowercase, {start => $start, end => $end};
+ }
+ elsif ($property eq "Uppercase") {
+ push @uppercase, {start => $start, end => $end};
+ }
+}
+
+my $num_category_ranges = scalar @category_ranges;
+my $num_alphabetic_ranges = scalar @alphabetic;
+my $num_lowercase_ranges = scalar @lowercase;
+my $num_uppercase_ranges = scalar @uppercase;
+my $num_white_space_ranges = scalar @white_space;
+my $num_hex_digit_ranges = scalar @hex_digits;
+my $num_join_control_ranges = scalar @join_control;
+
+# Start writing out the output file
open my $OT, '>', $output_table_file
or die "Could not open output file $output_table_file: $!\n";
-print $OT <<HEADER;
+print $OT <<"HEADER";
/*-------------------------------------------------------------------------
*
* unicode_category_table.h
@@ -188,11 +273,20 @@ typedef struct
uint8 category; /* General Category */
} pg_category_range;
-/* table of Unicode codepoint ranges and their categories */
-static const pg_category_range unicode_categories[$num_ranges] =
+typedef struct
{
+ uint32 first; /* Unicode codepoint */
+ uint32 last; /* Unicode codepoint */
+} pg_unicode_range;
+
HEADER
+print $OT <<"CATEGORY_TABLE";
+/* table of Unicode codepoint ranges and their categories */
+static const pg_category_range unicode_categories[$num_category_ranges] =
+{
+CATEGORY_TABLE
+
my $firsttime = 1;
foreach my $range (@category_ranges) {
printf $OT ",\n" unless $firsttime;
@@ -202,4 +296,101 @@ foreach my $range (@category_ranges) {
die "category missing: $range->{category}" unless $category;
printf $OT "\t{0x%06x, 0x%06x, %s}", $range->{start}, $range->{end}, $category;
}
+
+print $OT "\n};\n\n";
+
+print $OT <<"ALPHABETIC_TABLE";
+/* table of Unicode codepoint ranges of Alphabetic characters */
+static const pg_unicode_range unicode_alphabetic[$num_alphabetic_ranges] =
+{
+ALPHABETIC_TABLE
+
+$firsttime = 1;
+foreach my $range (@alphabetic) {
+ printf $OT ",\n" unless $firsttime;
+ $firsttime = 0;
+
+ printf $OT "\t{0x%06x, 0x%06x}", $range->{start}, $range->{end};
+}
+
+print $OT "\n};\n\n";
+
+print $OT <<"LOWERCASE_TABLE";
+/* table of Unicode codepoint ranges of Lowercase characters */
+static const pg_unicode_range unicode_lowercase[$num_lowercase_ranges] =
+{
+LOWERCASE_TABLE
+
+$firsttime = 1;
+foreach my $range (@lowercase) {
+ printf $OT ",\n" unless $firsttime;
+ $firsttime = 0;
+
+ printf $OT "\t{0x%06x, 0x%06x}", $range->{start}, $range->{end};
+}
+
+print $OT "\n};\n\n";
+
+print $OT <<"UPPERCASE_TABLE";
+/* table of Unicode codepoint ranges of Uppercase characters */
+static const pg_unicode_range unicode_uppercase[$num_uppercase_ranges] =
+{
+UPPERCASE_TABLE
+
+$firsttime = 1;
+foreach my $range (@uppercase) {
+ printf $OT ",\n" unless $firsttime;
+ $firsttime = 0;
+
+ printf $OT "\t{0x%06x, 0x%06x}", $range->{start}, $range->{end};
+}
+
+print $OT "\n};\n\n";
+
+print $OT <<"WHITE_SPACE_TABLE";
+/* table of Unicode codepoint ranges of White_Space characters */
+static const pg_unicode_range unicode_white_space[$num_white_space_ranges] =
+{
+WHITE_SPACE_TABLE
+
+$firsttime = 1;
+foreach my $range (@white_space) {
+ printf $OT ",\n" unless $firsttime;
+ $firsttime = 0;
+
+ printf $OT "\t{0x%06x, 0x%06x}", $range->{start}, $range->{end};
+}
+
+print $OT "\n};\n\n";
+
+print $OT <<"HEX_DIGITS_TABLE";
+/* table of Unicode codepoint ranges of Hex_Digit characters */
+static const pg_unicode_range unicode_hex_digit[$num_hex_digit_ranges] =
+{
+HEX_DIGITS_TABLE
+
+$firsttime = 1;
+foreach my $range (@hex_digits) {
+ printf $OT ",\n" unless $firsttime;
+ $firsttime = 0;
+
+ printf $OT "\t{0x%06x, 0x%06x}", $range->{start}, $range->{end};
+}
+
+print $OT "\n};\n\n";
+
+print $OT <<"JOIN_CONTROL_TABLE";
+/* table of Unicode codepoint ranges of Join_Control characters */
+static const pg_unicode_range unicode_join_control[$num_join_control_ranges] =
+{
+JOIN_CONTROL_TABLE
+
+$firsttime = 1;
+foreach my $range (@join_control) {
+ printf $OT ",\n" unless $firsttime;
+ $firsttime = 0;
+
+ printf $OT "\t{0x%06x, 0x%06x}", $range->{start}, $range->{end};
+}
+
print $OT "\n};\n";
diff --git a/src/common/unicode/meson.build b/src/common/unicode/meson.build
index e8cfdc1df4..3526ddb846 100644
--- a/src/common/unicode/meson.build
+++ b/src/common/unicode/meson.build
@@ -11,7 +11,7 @@ endif
# These files are part of the Unicode Character Database. Download them on
# demand.
-foreach f : ['CompositionExclusions.txt', 'DerivedNormalizationProps.txt', 'EastAsianWidth.txt', 'NormalizationTest.txt', 'UnicodeData.txt']
+foreach f : ['CompositionExclusions.txt', 'DerivedCoreProperties.txt', 'DerivedNormalizationProps.txt', 'EastAsianWidth.txt', 'NormalizationTest.txt', 'PropList.txt', 'UnicodeData.txt']
url = unicode_baseurl.format(UNICODE_VERSION, f)
target = custom_target(f,
output: f,
@@ -26,7 +26,7 @@ update_unicode_targets = []
update_unicode_targets += \
custom_target('unicode_category_table.h',
- input: [unicode_data['UnicodeData.txt']],
+ input: [unicode_data['UnicodeData.txt'], unicode_data['DerivedCoreProperties.txt'], unicode_data['PropList.txt']],
output: ['unicode_category_table.h'],
command: [
perl, files('generate-unicode_category_table.pl'),
diff --git a/src/common/unicode_category.c b/src/common/unicode_category.c
index 189cd6eca3..0b7a9947cc 100644
--- a/src/common/unicode_category.c
+++ b/src/common/unicode_category.c
@@ -1,6 +1,8 @@
/*-------------------------------------------------------------------------
* unicode_category.c
- * Determine general category of Unicode characters.
+ * Determine general category and character class of Unicode
+ * characters. Encoding must be UTF8, where we assume that the pg_wchar
+ * representation is a code point.
*
* Portions Copyright (c) 2017-2023, PostgreSQL Global Development Group
*
@@ -18,24 +20,78 @@
#include "common/unicode_category.h"
#include "common/unicode_category_table.h"
+/*
+ * We use a mask word for convenience when testing for multiple categories at
+ * once. The number of Unicode General Categories should never grow, so a
+ * 32-bit mask is fine.
+ */
+#define PG_U_CATEGORY_MASK(X) ((uint32)(1 << (X)))
+
+#define PG_U_LU_MASK PG_U_CATEGORY_MASK(PG_U_UPPERCASE_LETTER)
+#define PG_U_LL_MASK PG_U_CATEGORY_MASK(PG_U_LOWERCASE_LETTER)
+#define PG_U_LT_MASK PG_U_CATEGORY_MASK(PG_U_TITLECASE_LETTER)
+#define PG_U_LC_MASK (PG_U_LU_MASK|PG_U_LL_MASK|PG_U_LT_MASK)
+#define PG_U_LM_MASK PG_U_CATEGORY_MASK(PG_U_MODIFIER_LETTER)
+#define PG_U_LO_MASK PG_U_CATEGORY_MASK(PG_U_OTHER_LETTER)
+#define PG_U_L_MASK (PG_U_LU_MASK|PG_U_LL_MASK|PG_U_LT_MASK|PG_U_LM_MASK|\
+ PG_U_LO_MASK)
+#define PG_U_MN_MASK PG_U_CATEGORY_MASK(PG_U_NONSPACING_MARK)
+#define PG_U_ME_MASK PG_U_CATEGORY_MASK(PG_U_ENCLOSING_MARK)
+#define PG_U_MC_MASK PG_U_CATEGORY_MASK(PG_U_SPACING_MARK)
+#define PG_U_M_MASK (PG_U_MN_MASK|PG_U_MC_MASK|PG_U_ME_MASK)
+#define PG_U_ND_MASK PG_U_CATEGORY_MASK(PG_U_DECIMAL_NUMBER)
+#define PG_U_NL_MASK PG_U_CATEGORY_MASK(PG_U_LETTER_NUMBER)
+#define PG_U_NO_MASK PG_U_CATEGORY_MASK(PG_U_OTHER_NUMBER)
+#define PG_U_N_MASK (PG_U_ND_MASK|PG_U_NL_MASK|PG_U_NO_MASK)
+#define PG_U_PC_MASK PG_U_CATEGORY_MASK(PG_U_CONNECTOR_PUNCTUATION)
+#define PG_U_PD_MASK PG_U_CATEGORY_MASK(PG_U_DASH_PUNCTUATION)
+#define PG_U_PS_MASK PG_U_CATEGORY_MASK(PG_U_OPEN_PUNCTUATION)
+#define PG_U_PE_MASK PG_U_CATEGORY_MASK(PG_U_CLOSE_PUNCTUATION)
+#define PG_U_PI_MASK PG_U_CATEGORY_MASK(PG_U_INITIAL_PUNCTUATION)
+#define PG_U_PF_MASK PG_U_CATEGORY_MASK(PG_U_FINAL_PUNCTUATION)
+#define PG_U_PO_MASK PG_U_CATEGORY_MASK(PG_U_OTHER_PUNCTUATION)
+#define PG_U_P_MASK (PG_U_PC_MASK|PG_U_PD_MASK|PG_U_PS_MASK|PG_U_PE_MASK|\
+ PG_U_PI_MASK|PG_U_PF_MASK|PG_U_PO_MASK)
+#define PG_U_SM_MASK PG_U_CATEGORY_MASK(PG_U_MATH_SYMBOL)
+#define PG_U_SC_MASK PG_U_CATEGORY_MASK(PG_U_CURRENCY_SYMBOL)
+#define PG_U_SK_MASK PG_U_CATEGORY_MASK(PG_U_MODIFIER_SYMBOL)
+#define PG_U_SO_MASK PG_U_CATEGORY_MASK(PG_U_OTHER_SYMBOL)
+#define PG_U_S_MASK (PG_U_SM_MASK|PG_U_SC_MASK|PG_U_SK_MASK|PG_U_SO_MASK)
+#define PG_U_ZS_MASK PG_U_CATEGORY_MASK(PG_U_SPACE_SEPARATOR)
+#define PG_U_ZL_MASK PG_U_CATEGORY_MASK(PG_U_LINE_SEPARATOR)
+#define PG_U_ZP_MASK PG_U_CATEGORY_MASK(PG_U_PARAGRAPH_SEPARATOR)
+#define PG_U_Z_MASK (PG_U_ZS_MASK|PG_U_ZL_MASK|PG_U_ZP_MASK)
+#define PG_U_CC_MASK PG_U_CATEGORY_MASK(PG_U_CONTROL)
+#define PG_U_CF_MASK PG_U_CATEGORY_MASK(PG_U_FORMAT)
+#define PG_U_CS_MASK PG_U_CATEGORY_MASK(PG_U_SURROGATE)
+#define PG_U_CO_MASK PG_U_CATEGORY_MASK(PG_U_PRIVATE_USE)
+#define PG_U_CN_MASK PG_U_CATEGORY_MASK(PG_U_UNASSIGNED)
+#define PG_U_C_MASK (PG_U_CC_MASK|PG_U_CF_MASK|PG_U_CS_MASK|PG_U_CO_MASK|\
+ PG_U_CN_MASK)
+
+#define PG_U_CHARACTER_TAB 0x09
+
+static bool range_search(const pg_unicode_range * tbl, Size size,
+ pg_wchar code);
+
/*
* Unicode general category for the given codepoint.
*/
pg_unicode_category
-unicode_category(pg_wchar ucs)
+unicode_category(pg_wchar code)
{
int min = 0;
int mid;
int max = lengthof(unicode_categories) - 1;
- Assert(ucs <= 0x10ffff);
+ Assert(code <= 0x10ffff);
while (max >= min)
{
mid = (min + max) / 2;
- if (ucs > unicode_categories[mid].last)
+ if (code > unicode_categories[mid].last)
min = mid + 1;
- else if (ucs < unicode_categories[mid].first)
+ else if (code < unicode_categories[mid].first)
max = mid - 1;
else
return unicode_categories[mid].category;
@@ -44,6 +100,123 @@ unicode_category(pg_wchar ucs)
return PG_U_UNASSIGNED;
}
+/*
+ * The following functions implement the regex character classification as
+ * described at: http://www.unicode.org/reports/tr18/#Compatibility_Properties
+ */
+
+bool
+pg_u_isdigit(pg_wchar code)
+{
+ return unicode_category(code) == PG_U_DECIMAL_NUMBER;
+}
+
+bool
+pg_u_isalpha(pg_wchar code)
+{
+ return range_search(unicode_alphabetic, lengthof(unicode_alphabetic),
+ code);
+}
+
+bool
+pg_u_isalnum(pg_wchar code)
+{
+ return pg_u_isalpha(code) || pg_u_isdigit(code);
+}
+
+bool
+pg_u_isword(pg_wchar code)
+{
+ uint32 category_mask = PG_U_CATEGORY_MASK(unicode_category(code));
+
+ return
+ category_mask & (PG_U_M_MASK|PG_U_ND_MASK|PG_U_PC_MASK) ||
+ pg_u_isalpha(code) ||
+ range_search(unicode_join_control, lengthof(unicode_join_control),
+ code);
+}
+
+bool
+pg_u_isupper(pg_wchar code)
+{
+ return range_search(unicode_uppercase, lengthof(unicode_uppercase), code);
+}
+
+bool
+pg_u_islower(pg_wchar code)
+{
+ return range_search(unicode_lowercase, lengthof(unicode_lowercase), code);
+}
+
+bool
+pg_u_isblank(pg_wchar code)
+{
+ return code == PG_U_CHARACTER_TAB ||
+ unicode_category(code) == PG_U_SPACE_SEPARATOR;
+}
+
+bool
+pg_u_iscntrl(pg_wchar code)
+{
+ return unicode_category(code) == PG_U_CONTROL;
+}
+
+bool
+pg_u_isgraph(pg_wchar code)
+{
+ uint32 category_mask = PG_U_CATEGORY_MASK(unicode_category(code));
+
+ if (category_mask & (PG_U_CC_MASK|PG_U_CS_MASK|PG_U_CN_MASK) ||
+ pg_u_isspace(code))
+ return false;
+ return true;
+}
+
+bool
+pg_u_isprint(pg_wchar code)
+{
+ pg_unicode_category category = unicode_category(code);
+
+ if (category == PG_U_CONTROL)
+ return false;
+
+ return pg_u_isgraph(code) || pg_u_isblank(code);
+}
+
+bool
+pg_u_ispunct(pg_wchar code)
+{
+ uint32 category_mask = PG_U_CATEGORY_MASK(unicode_category(code));
+
+ return category_mask & PG_U_P_MASK;
+}
+
+/* posix variant */
+bool
+pg_u_ispunct_posix(pg_wchar code)
+{
+ uint32 category_mask;
+
+ if (pg_u_isalpha(code))
+ return false;
+
+ category_mask = PG_U_CATEGORY_MASK(unicode_category(code));
+ return category_mask & (PG_U_P_MASK|PG_U_S_MASK);
+}
+
+bool
+pg_u_isspace(pg_wchar code)
+{
+ return range_search(unicode_white_space, lengthof(unicode_white_space),
+ code);
+}
+
+bool
+pg_u_isxdigit(pg_wchar code)
+{
+ return range_search(unicode_hex_digit, lengthof(unicode_hex_digit), code);
+}
+
/*
* Description of Unicode general category.
*/
@@ -191,3 +364,30 @@ unicode_category_abbrev(pg_unicode_category category)
Assert(false);
return "??"; /* keep compiler quiet */
}
+
+/*
+ * Binary search to test if given codepoint exists in one of the ranges in the
+ * given table.
+ */
+static bool
+range_search(const pg_unicode_range * tbl, Size size, pg_wchar code)
+{
+ int min = 0;
+ int mid;
+ int max = size - 1;
+
+ Assert(code <= 0x10ffff);
+
+ while (max >= min)
+ {
+ mid = (min + max) / 2;
+ if (code > tbl[mid].last)
+ min = mid + 1;
+ else if (code < tbl[mid].first)
+ max = mid - 1;
+ else
+ return true;
+ }
+
+ return false;
+}
diff --git a/src/include/common/unicode_category.h b/src/include/common/unicode_category.h
index 81d38c7411..25c7937a43 100644
--- a/src/include/common/unicode_category.h
+++ b/src/include/common/unicode_category.h
@@ -62,7 +62,22 @@ typedef enum pg_unicode_category
} pg_unicode_category;
extern pg_unicode_category unicode_category(pg_wchar ucs);
-const char *unicode_category_string(pg_unicode_category category);
-const char *unicode_category_abbrev(pg_unicode_category category);
+extern const char *unicode_category_string(pg_unicode_category category);
+extern const char *unicode_category_abbrev(pg_unicode_category category);
+
+extern bool pg_u_isdigit(pg_wchar c);
+extern bool pg_u_isalpha(pg_wchar c);
+extern bool pg_u_isalnum(pg_wchar c);
+extern bool pg_u_isword(pg_wchar c);
+extern bool pg_u_isupper(pg_wchar c);
+extern bool pg_u_islower(pg_wchar c);
+extern bool pg_u_isblank(pg_wchar c);
+extern bool pg_u_iscntrl(pg_wchar c);
+extern bool pg_u_isgraph(pg_wchar c);
+extern bool pg_u_isprint(pg_wchar c);
+extern bool pg_u_ispunct(pg_wchar c);
+extern bool pg_u_ispunct_posix(pg_wchar c);
+extern bool pg_u_isspace(pg_wchar c);
+extern bool pg_u_isxdigit(pg_wchar c);
#endif /* UNICODE_CATEGORY_H */
diff --git a/src/include/common/unicode_category_table.h b/src/include/common/unicode_category_table.h
index 14f1ea0677..86cdc9c0ed 100644
--- a/src/include/common/unicode_category_table.h
+++ b/src/include/common/unicode_category_table.h
@@ -25,6 +25,12 @@ typedef struct
uint8 category; /* General Category */
} pg_category_range;
+typedef struct
+{
+ uint32 first; /* Unicode codepoint */
+ uint32 last; /* Unicode codepoint */
+} pg_unicode_range;
+
/* table of Unicode codepoint ranges and their categories */
static const pg_category_range unicode_categories[3302] =
{
@@ -3331,3 +3337,2529 @@ static const pg_category_range unicode_categories[3302] =
{0x0f0000, 0x0ffffd, PG_U_PRIVATE_USE},
{0x100000, 0x10fffd, PG_U_PRIVATE_USE}
};
+
+/* table of Unicode codepoint ranges of Alphabetic characters */
+static const pg_unicode_range unicode_alphabetic[1141] =
+{
+ {0x000041, 0x00005a},
+ {0x000061, 0x00007a},
+ {0x0000aa, 0x0000aa},
+ {0x0000b5, 0x0000b5},
+ {0x0000ba, 0x0000ba},
+ {0x0000c0, 0x0000d6},
+ {0x0000d8, 0x0000f6},
+ {0x0000f8, 0x0001ba},
+ {0x0001bb, 0x0001bb},
+ {0x0001bc, 0x0001bf},
+ {0x0001c0, 0x0001c3},
+ {0x0001c4, 0x000293},
+ {0x000294, 0x000294},
+ {0x000295, 0x0002af},
+ {0x0002b0, 0x0002c1},
+ {0x0002c6, 0x0002d1},
+ {0x0002e0, 0x0002e4},
+ {0x0002ec, 0x0002ec},
+ {0x0002ee, 0x0002ee},
+ {0x000345, 0x000345},
+ {0x000370, 0x000373},
+ {0x000374, 0x000374},
+ {0x000376, 0x000377},
+ {0x00037a, 0x00037a},
+ {0x00037b, 0x00037d},
+ {0x00037f, 0x00037f},
+ {0x000386, 0x000386},
+ {0x000388, 0x00038a},
+ {0x00038c, 0x00038c},
+ {0x00038e, 0x0003a1},
+ {0x0003a3, 0x0003f5},
+ {0x0003f7, 0x000481},
+ {0x00048a, 0x00052f},
+ {0x000531, 0x000556},
+ {0x000559, 0x000559},
+ {0x000560, 0x000588},
+ {0x0005b0, 0x0005bd},
+ {0x0005bf, 0x0005bf},
+ {0x0005c1, 0x0005c2},
+ {0x0005c4, 0x0005c5},
+ {0x0005c7, 0x0005c7},
+ {0x0005d0, 0x0005ea},
+ {0x0005ef, 0x0005f2},
+ {0x000610, 0x00061a},
+ {0x000620, 0x00063f},
+ {0x000640, 0x000640},
+ {0x000641, 0x00064a},
+ {0x00064b, 0x000657},
+ {0x000659, 0x00065f},
+ {0x00066e, 0x00066f},
+ {0x000670, 0x000670},
+ {0x000671, 0x0006d3},
+ {0x0006d5, 0x0006d5},
+ {0x0006d6, 0x0006dc},
+ {0x0006e1, 0x0006e4},
+ {0x0006e5, 0x0006e6},
+ {0x0006e7, 0x0006e8},
+ {0x0006ed, 0x0006ed},
+ {0x0006ee, 0x0006ef},
+ {0x0006fa, 0x0006fc},
+ {0x0006ff, 0x0006ff},
+ {0x000710, 0x000710},
+ {0x000711, 0x000711},
+ {0x000712, 0x00072f},
+ {0x000730, 0x00073f},
+ {0x00074d, 0x0007a5},
+ {0x0007a6, 0x0007b0},
+ {0x0007b1, 0x0007b1},
+ {0x0007ca, 0x0007ea},
+ {0x0007f4, 0x0007f5},
+ {0x0007fa, 0x0007fa},
+ {0x000800, 0x000815},
+ {0x000816, 0x000817},
+ {0x00081a, 0x00081a},
+ {0x00081b, 0x000823},
+ {0x000824, 0x000824},
+ {0x000825, 0x000827},
+ {0x000828, 0x000828},
+ {0x000829, 0x00082c},
+ {0x000840, 0x000858},
+ {0x000860, 0x00086a},
+ {0x000870, 0x000887},
+ {0x000889, 0x00088e},
+ {0x0008a0, 0x0008c8},
+ {0x0008c9, 0x0008c9},
+ {0x0008d4, 0x0008df},
+ {0x0008e3, 0x0008e9},
+ {0x0008f0, 0x000902},
+ {0x000903, 0x000903},
+ {0x000904, 0x000939},
+ {0x00093a, 0x00093a},
+ {0x00093b, 0x00093b},
+ {0x00093d, 0x00093d},
+ {0x00093e, 0x000940},
+ {0x000941, 0x000948},
+ {0x000949, 0x00094c},
+ {0x00094e, 0x00094f},
+ {0x000950, 0x000950},
+ {0x000955, 0x000957},
+ {0x000958, 0x000961},
+ {0x000962, 0x000963},
+ {0x000971, 0x000971},
+ {0x000972, 0x000980},
+ {0x000981, 0x000981},
+ {0x000982, 0x000983},
+ {0x000985, 0x00098c},
+ {0x00098f, 0x000990},
+ {0x000993, 0x0009a8},
+ {0x0009aa, 0x0009b0},
+ {0x0009b2, 0x0009b2},
+ {0x0009b6, 0x0009b9},
+ {0x0009bd, 0x0009bd},
+ {0x0009be, 0x0009c0},
+ {0x0009c1, 0x0009c4},
+ {0x0009c7, 0x0009c8},
+ {0x0009cb, 0x0009cc},
+ {0x0009ce, 0x0009ce},
+ {0x0009d7, 0x0009d7},
+ {0x0009dc, 0x0009dd},
+ {0x0009df, 0x0009e1},
+ {0x0009e2, 0x0009e3},
+ {0x0009f0, 0x0009f1},
+ {0x0009fc, 0x0009fc},
+ {0x000a01, 0x000a02},
+ {0x000a03, 0x000a03},
+ {0x000a05, 0x000a0a},
+ {0x000a0f, 0x000a10},
+ {0x000a13, 0x000a28},
+ {0x000a2a, 0x000a30},
+ {0x000a32, 0x000a33},
+ {0x000a35, 0x000a36},
+ {0x000a38, 0x000a39},
+ {0x000a3e, 0x000a40},
+ {0x000a41, 0x000a42},
+ {0x000a47, 0x000a48},
+ {0x000a4b, 0x000a4c},
+ {0x000a51, 0x000a51},
+ {0x000a59, 0x000a5c},
+ {0x000a5e, 0x000a5e},
+ {0x000a70, 0x000a71},
+ {0x000a72, 0x000a74},
+ {0x000a75, 0x000a75},
+ {0x000a81, 0x000a82},
+ {0x000a83, 0x000a83},
+ {0x000a85, 0x000a8d},
+ {0x000a8f, 0x000a91},
+ {0x000a93, 0x000aa8},
+ {0x000aaa, 0x000ab0},
+ {0x000ab2, 0x000ab3},
+ {0x000ab5, 0x000ab9},
+ {0x000abd, 0x000abd},
+ {0x000abe, 0x000ac0},
+ {0x000ac1, 0x000ac5},
+ {0x000ac7, 0x000ac8},
+ {0x000ac9, 0x000ac9},
+ {0x000acb, 0x000acc},
+ {0x000ad0, 0x000ad0},
+ {0x000ae0, 0x000ae1},
+ {0x000ae2, 0x000ae3},
+ {0x000af9, 0x000af9},
+ {0x000afa, 0x000afc},
+ {0x000b01, 0x000b01},
+ {0x000b02, 0x000b03},
+ {0x000b05, 0x000b0c},
+ {0x000b0f, 0x000b10},
+ {0x000b13, 0x000b28},
+ {0x000b2a, 0x000b30},
+ {0x000b32, 0x000b33},
+ {0x000b35, 0x000b39},
+ {0x000b3d, 0x000b3d},
+ {0x000b3e, 0x000b3e},
+ {0x000b3f, 0x000b3f},
+ {0x000b40, 0x000b40},
+ {0x000b41, 0x000b44},
+ {0x000b47, 0x000b48},
+ {0x000b4b, 0x000b4c},
+ {0x000b56, 0x000b56},
+ {0x000b57, 0x000b57},
+ {0x000b5c, 0x000b5d},
+ {0x000b5f, 0x000b61},
+ {0x000b62, 0x000b63},
+ {0x000b71, 0x000b71},
+ {0x000b82, 0x000b82},
+ {0x000b83, 0x000b83},
+ {0x000b85, 0x000b8a},
+ {0x000b8e, 0x000b90},
+ {0x000b92, 0x000b95},
+ {0x000b99, 0x000b9a},
+ {0x000b9c, 0x000b9c},
+ {0x000b9e, 0x000b9f},
+ {0x000ba3, 0x000ba4},
+ {0x000ba8, 0x000baa},
+ {0x000bae, 0x000bb9},
+ {0x000bbe, 0x000bbf},
+ {0x000bc0, 0x000bc0},
+ {0x000bc1, 0x000bc2},
+ {0x000bc6, 0x000bc8},
+ {0x000bca, 0x000bcc},
+ {0x000bd0, 0x000bd0},
+ {0x000bd7, 0x000bd7},
+ {0x000c00, 0x000c00},
+ {0x000c01, 0x000c03},
+ {0x000c04, 0x000c04},
+ {0x000c05, 0x000c0c},
+ {0x000c0e, 0x000c10},
+ {0x000c12, 0x000c28},
+ {0x000c2a, 0x000c39},
+ {0x000c3d, 0x000c3d},
+ {0x000c3e, 0x000c40},
+ {0x000c41, 0x000c44},
+ {0x000c46, 0x000c48},
+ {0x000c4a, 0x000c4c},
+ {0x000c55, 0x000c56},
+ {0x000c58, 0x000c5a},
+ {0x000c5d, 0x000c5d},
+ {0x000c60, 0x000c61},
+ {0x000c62, 0x000c63},
+ {0x000c80, 0x000c80},
+ {0x000c81, 0x000c81},
+ {0x000c82, 0x000c83},
+ {0x000c85, 0x000c8c},
+ {0x000c8e, 0x000c90},
+ {0x000c92, 0x000ca8},
+ {0x000caa, 0x000cb3},
+ {0x000cb5, 0x000cb9},
+ {0x000cbd, 0x000cbd},
+ {0x000cbe, 0x000cbe},
+ {0x000cbf, 0x000cbf},
+ {0x000cc0, 0x000cc4},
+ {0x000cc6, 0x000cc6},
+ {0x000cc7, 0x000cc8},
+ {0x000cca, 0x000ccb},
+ {0x000ccc, 0x000ccc},
+ {0x000cd5, 0x000cd6},
+ {0x000cdd, 0x000cde},
+ {0x000ce0, 0x000ce1},
+ {0x000ce2, 0x000ce3},
+ {0x000cf1, 0x000cf2},
+ {0x000cf3, 0x000cf3},
+ {0x000d00, 0x000d01},
+ {0x000d02, 0x000d03},
+ {0x000d04, 0x000d0c},
+ {0x000d0e, 0x000d10},
+ {0x000d12, 0x000d3a},
+ {0x000d3d, 0x000d3d},
+ {0x000d3e, 0x000d40},
+ {0x000d41, 0x000d44},
+ {0x000d46, 0x000d48},
+ {0x000d4a, 0x000d4c},
+ {0x000d4e, 0x000d4e},
+ {0x000d54, 0x000d56},
+ {0x000d57, 0x000d57},
+ {0x000d5f, 0x000d61},
+ {0x000d62, 0x000d63},
+ {0x000d7a, 0x000d7f},
+ {0x000d81, 0x000d81},
+ {0x000d82, 0x000d83},
+ {0x000d85, 0x000d96},
+ {0x000d9a, 0x000db1},
+ {0x000db3, 0x000dbb},
+ {0x000dbd, 0x000dbd},
+ {0x000dc0, 0x000dc6},
+ {0x000dcf, 0x000dd1},
+ {0x000dd2, 0x000dd4},
+ {0x000dd6, 0x000dd6},
+ {0x000dd8, 0x000ddf},
+ {0x000df2, 0x000df3},
+ {0x000e01, 0x000e30},
+ {0x000e31, 0x000e31},
+ {0x000e32, 0x000e33},
+ {0x000e34, 0x000e3a},
+ {0x000e40, 0x000e45},
+ {0x000e46, 0x000e46},
+ {0x000e4d, 0x000e4d},
+ {0x000e81, 0x000e82},
+ {0x000e84, 0x000e84},
+ {0x000e86, 0x000e8a},
+ {0x000e8c, 0x000ea3},
+ {0x000ea5, 0x000ea5},
+ {0x000ea7, 0x000eb0},
+ {0x000eb1, 0x000eb1},
+ {0x000eb2, 0x000eb3},
+ {0x000eb4, 0x000eb9},
+ {0x000ebb, 0x000ebc},
+ {0x000ebd, 0x000ebd},
+ {0x000ec0, 0x000ec4},
+ {0x000ec6, 0x000ec6},
+ {0x000ecd, 0x000ecd},
+ {0x000edc, 0x000edf},
+ {0x000f00, 0x000f00},
+ {0x000f40, 0x000f47},
+ {0x000f49, 0x000f6c},
+ {0x000f71, 0x000f7e},
+ {0x000f7f, 0x000f7f},
+ {0x000f80, 0x000f83},
+ {0x000f88, 0x000f8c},
+ {0x000f8d, 0x000f97},
+ {0x000f99, 0x000fbc},
+ {0x001000, 0x00102a},
+ {0x00102b, 0x00102c},
+ {0x00102d, 0x001030},
+ {0x001031, 0x001031},
+ {0x001032, 0x001036},
+ {0x001038, 0x001038},
+ {0x00103b, 0x00103c},
+ {0x00103d, 0x00103e},
+ {0x00103f, 0x00103f},
+ {0x001050, 0x001055},
+ {0x001056, 0x001057},
+ {0x001058, 0x001059},
+ {0x00105a, 0x00105d},
+ {0x00105e, 0x001060},
+ {0x001061, 0x001061},
+ {0x001062, 0x001064},
+ {0x001065, 0x001066},
+ {0x001067, 0x00106d},
+ {0x00106e, 0x001070},
+ {0x001071, 0x001074},
+ {0x001075, 0x001081},
+ {0x001082, 0x001082},
+ {0x001083, 0x001084},
+ {0x001085, 0x001086},
+ {0x001087, 0x00108c},
+ {0x00108d, 0x00108d},
+ {0x00108e, 0x00108e},
+ {0x00108f, 0x00108f},
+ {0x00109a, 0x00109c},
+ {0x00109d, 0x00109d},
+ {0x0010a0, 0x0010c5},
+ {0x0010c7, 0x0010c7},
+ {0x0010cd, 0x0010cd},
+ {0x0010d0, 0x0010fa},
+ {0x0010fc, 0x0010fc},
+ {0x0010fd, 0x0010ff},
+ {0x001100, 0x001248},
+ {0x00124a, 0x00124d},
+ {0x001250, 0x001256},
+ {0x001258, 0x001258},
+ {0x00125a, 0x00125d},
+ {0x001260, 0x001288},
+ {0x00128a, 0x00128d},
+ {0x001290, 0x0012b0},
+ {0x0012b2, 0x0012b5},
+ {0x0012b8, 0x0012be},
+ {0x0012c0, 0x0012c0},
+ {0x0012c2, 0x0012c5},
+ {0x0012c8, 0x0012d6},
+ {0x0012d8, 0x001310},
+ {0x001312, 0x001315},
+ {0x001318, 0x00135a},
+ {0x001380, 0x00138f},
+ {0x0013a0, 0x0013f5},
+ {0x0013f8, 0x0013fd},
+ {0x001401, 0x00166c},
+ {0x00166f, 0x00167f},
+ {0x001681, 0x00169a},
+ {0x0016a0, 0x0016ea},
+ {0x0016ee, 0x0016f0},
+ {0x0016f1, 0x0016f8},
+ {0x001700, 0x001711},
+ {0x001712, 0x001713},
+ {0x00171f, 0x001731},
+ {0x001732, 0x001733},
+ {0x001740, 0x001751},
+ {0x001752, 0x001753},
+ {0x001760, 0x00176c},
+ {0x00176e, 0x001770},
+ {0x001772, 0x001773},
+ {0x001780, 0x0017b3},
+ {0x0017b6, 0x0017b6},
+ {0x0017b7, 0x0017bd},
+ {0x0017be, 0x0017c5},
+ {0x0017c6, 0x0017c6},
+ {0x0017c7, 0x0017c8},
+ {0x0017d7, 0x0017d7},
+ {0x0017dc, 0x0017dc},
+ {0x001820, 0x001842},
+ {0x001843, 0x001843},
+ {0x001844, 0x001878},
+ {0x001880, 0x001884},
+ {0x001885, 0x001886},
+ {0x001887, 0x0018a8},
+ {0x0018a9, 0x0018a9},
+ {0x0018aa, 0x0018aa},
+ {0x0018b0, 0x0018f5},
+ {0x001900, 0x00191e},
+ {0x001920, 0x001922},
+ {0x001923, 0x001926},
+ {0x001927, 0x001928},
+ {0x001929, 0x00192b},
+ {0x001930, 0x001931},
+ {0x001932, 0x001932},
+ {0x001933, 0x001938},
+ {0x001950, 0x00196d},
+ {0x001970, 0x001974},
+ {0x001980, 0x0019ab},
+ {0x0019b0, 0x0019c9},
+ {0x001a00, 0x001a16},
+ {0x001a17, 0x001a18},
+ {0x001a19, 0x001a1a},
+ {0x001a1b, 0x001a1b},
+ {0x001a20, 0x001a54},
+ {0x001a55, 0x001a55},
+ {0x001a56, 0x001a56},
+ {0x001a57, 0x001a57},
+ {0x001a58, 0x001a5e},
+ {0x001a61, 0x001a61},
+ {0x001a62, 0x001a62},
+ {0x001a63, 0x001a64},
+ {0x001a65, 0x001a6c},
+ {0x001a6d, 0x001a72},
+ {0x001a73, 0x001a74},
+ {0x001aa7, 0x001aa7},
+ {0x001abf, 0x001ac0},
+ {0x001acc, 0x001ace},
+ {0x001b00, 0x001b03},
+ {0x001b04, 0x001b04},
+ {0x001b05, 0x001b33},
+ {0x001b35, 0x001b35},
+ {0x001b36, 0x001b3a},
+ {0x001b3b, 0x001b3b},
+ {0x001b3c, 0x001b3c},
+ {0x001b3d, 0x001b41},
+ {0x001b42, 0x001b42},
+ {0x001b43, 0x001b43},
+ {0x001b45, 0x001b4c},
+ {0x001b80, 0x001b81},
+ {0x001b82, 0x001b82},
+ {0x001b83, 0x001ba0},
+ {0x001ba1, 0x001ba1},
+ {0x001ba2, 0x001ba5},
+ {0x001ba6, 0x001ba7},
+ {0x001ba8, 0x001ba9},
+ {0x001bac, 0x001bad},
+ {0x001bae, 0x001baf},
+ {0x001bba, 0x001be5},
+ {0x001be7, 0x001be7},
+ {0x001be8, 0x001be9},
+ {0x001bea, 0x001bec},
+ {0x001bed, 0x001bed},
+ {0x001bee, 0x001bee},
+ {0x001bef, 0x001bf1},
+ {0x001c00, 0x001c23},
+ {0x001c24, 0x001c2b},
+ {0x001c2c, 0x001c33},
+ {0x001c34, 0x001c35},
+ {0x001c36, 0x001c36},
+ {0x001c4d, 0x001c4f},
+ {0x001c5a, 0x001c77},
+ {0x001c78, 0x001c7d},
+ {0x001c80, 0x001c88},
+ {0x001c90, 0x001cba},
+ {0x001cbd, 0x001cbf},
+ {0x001ce9, 0x001cec},
+ {0x001cee, 0x001cf3},
+ {0x001cf5, 0x001cf6},
+ {0x001cfa, 0x001cfa},
+ {0x001d00, 0x001d2b},
+ {0x001d2c, 0x001d6a},
+ {0x001d6b, 0x001d77},
+ {0x001d78, 0x001d78},
+ {0x001d79, 0x001d9a},
+ {0x001d9b, 0x001dbf},
+ {0x001de7, 0x001df4},
+ {0x001e00, 0x001f15},
+ {0x001f18, 0x001f1d},
+ {0x001f20, 0x001f45},
+ {0x001f48, 0x001f4d},
+ {0x001f50, 0x001f57},
+ {0x001f59, 0x001f59},
+ {0x001f5b, 0x001f5b},
+ {0x001f5d, 0x001f5d},
+ {0x001f5f, 0x001f7d},
+ {0x001f80, 0x001fb4},
+ {0x001fb6, 0x001fbc},
+ {0x001fbe, 0x001fbe},
+ {0x001fc2, 0x001fc4},
+ {0x001fc6, 0x001fcc},
+ {0x001fd0, 0x001fd3},
+ {0x001fd6, 0x001fdb},
+ {0x001fe0, 0x001fec},
+ {0x001ff2, 0x001ff4},
+ {0x001ff6, 0x001ffc},
+ {0x002071, 0x002071},
+ {0x00207f, 0x00207f},
+ {0x002090, 0x00209c},
+ {0x002102, 0x002102},
+ {0x002107, 0x002107},
+ {0x00210a, 0x002113},
+ {0x002115, 0x002115},
+ {0x002119, 0x00211d},
+ {0x002124, 0x002124},
+ {0x002126, 0x002126},
+ {0x002128, 0x002128},
+ {0x00212a, 0x00212d},
+ {0x00212f, 0x002134},
+ {0x002135, 0x002138},
+ {0x002139, 0x002139},
+ {0x00213c, 0x00213f},
+ {0x002145, 0x002149},
+ {0x00214e, 0x00214e},
+ {0x002160, 0x002182},
+ {0x002183, 0x002184},
+ {0x002185, 0x002188},
+ {0x0024b6, 0x0024e9},
+ {0x002c00, 0x002c7b},
+ {0x002c7c, 0x002c7d},
+ {0x002c7e, 0x002ce4},
+ {0x002ceb, 0x002cee},
+ {0x002cf2, 0x002cf3},
+ {0x002d00, 0x002d25},
+ {0x002d27, 0x002d27},
+ {0x002d2d, 0x002d2d},
+ {0x002d30, 0x002d67},
+ {0x002d6f, 0x002d6f},
+ {0x002d80, 0x002d96},
+ {0x002da0, 0x002da6},
+ {0x002da8, 0x002dae},
+ {0x002db0, 0x002db6},
+ {0x002db8, 0x002dbe},
+ {0x002dc0, 0x002dc6},
+ {0x002dc8, 0x002dce},
+ {0x002dd0, 0x002dd6},
+ {0x002dd8, 0x002dde},
+ {0x002de0, 0x002dff},
+ {0x002e2f, 0x002e2f},
+ {0x003005, 0x003005},
+ {0x003006, 0x003006},
+ {0x003007, 0x003007},
+ {0x003021, 0x003029},
+ {0x003031, 0x003035},
+ {0x003038, 0x00303a},
+ {0x00303b, 0x00303b},
+ {0x00303c, 0x00303c},
+ {0x003041, 0x003096},
+ {0x00309d, 0x00309e},
+ {0x00309f, 0x00309f},
+ {0x0030a1, 0x0030fa},
+ {0x0030fc, 0x0030fe},
+ {0x0030ff, 0x0030ff},
+ {0x003105, 0x00312f},
+ {0x003131, 0x00318e},
+ {0x0031a0, 0x0031bf},
+ {0x0031f0, 0x0031ff},
+ {0x003400, 0x004dbf},
+ {0x004e00, 0x00a014},
+ {0x00a015, 0x00a015},
+ {0x00a016, 0x00a48c},
+ {0x00a4d0, 0x00a4f7},
+ {0x00a4f8, 0x00a4fd},
+ {0x00a500, 0x00a60b},
+ {0x00a60c, 0x00a60c},
+ {0x00a610, 0x00a61f},
+ {0x00a62a, 0x00a62b},
+ {0x00a640, 0x00a66d},
+ {0x00a66e, 0x00a66e},
+ {0x00a674, 0x00a67b},
+ {0x00a67f, 0x00a67f},
+ {0x00a680, 0x00a69b},
+ {0x00a69c, 0x00a69d},
+ {0x00a69e, 0x00a69f},
+ {0x00a6a0, 0x00a6e5},
+ {0x00a6e6, 0x00a6ef},
+ {0x00a717, 0x00a71f},
+ {0x00a722, 0x00a76f},
+ {0x00a770, 0x00a770},
+ {0x00a771, 0x00a787},
+ {0x00a788, 0x00a788},
+ {0x00a78b, 0x00a78e},
+ {0x00a78f, 0x00a78f},
+ {0x00a790, 0x00a7ca},
+ {0x00a7d0, 0x00a7d1},
+ {0x00a7d3, 0x00a7d3},
+ {0x00a7d5, 0x00a7d9},
+ {0x00a7f2, 0x00a7f4},
+ {0x00a7f5, 0x00a7f6},
+ {0x00a7f7, 0x00a7f7},
+ {0x00a7f8, 0x00a7f9},
+ {0x00a7fa, 0x00a7fa},
+ {0x00a7fb, 0x00a801},
+ {0x00a802, 0x00a802},
+ {0x00a803, 0x00a805},
+ {0x00a807, 0x00a80a},
+ {0x00a80b, 0x00a80b},
+ {0x00a80c, 0x00a822},
+ {0x00a823, 0x00a824},
+ {0x00a825, 0x00a826},
+ {0x00a827, 0x00a827},
+ {0x00a840, 0x00a873},
+ {0x00a880, 0x00a881},
+ {0x00a882, 0x00a8b3},
+ {0x00a8b4, 0x00a8c3},
+ {0x00a8c5, 0x00a8c5},
+ {0x00a8f2, 0x00a8f7},
+ {0x00a8fb, 0x00a8fb},
+ {0x00a8fd, 0x00a8fe},
+ {0x00a8ff, 0x00a8ff},
+ {0x00a90a, 0x00a925},
+ {0x00a926, 0x00a92a},
+ {0x00a930, 0x00a946},
+ {0x00a947, 0x00a951},
+ {0x00a952, 0x00a952},
+ {0x00a960, 0x00a97c},
+ {0x00a980, 0x00a982},
+ {0x00a983, 0x00a983},
+ {0x00a984, 0x00a9b2},
+ {0x00a9b4, 0x00a9b5},
+ {0x00a9b6, 0x00a9b9},
+ {0x00a9ba, 0x00a9bb},
+ {0x00a9bc, 0x00a9bd},
+ {0x00a9be, 0x00a9bf},
+ {0x00a9cf, 0x00a9cf},
+ {0x00a9e0, 0x00a9e4},
+ {0x00a9e5, 0x00a9e5},
+ {0x00a9e6, 0x00a9e6},
+ {0x00a9e7, 0x00a9ef},
+ {0x00a9fa, 0x00a9fe},
+ {0x00aa00, 0x00aa28},
+ {0x00aa29, 0x00aa2e},
+ {0x00aa2f, 0x00aa30},
+ {0x00aa31, 0x00aa32},
+ {0x00aa33, 0x00aa34},
+ {0x00aa35, 0x00aa36},
+ {0x00aa40, 0x00aa42},
+ {0x00aa43, 0x00aa43},
+ {0x00aa44, 0x00aa4b},
+ {0x00aa4c, 0x00aa4c},
+ {0x00aa4d, 0x00aa4d},
+ {0x00aa60, 0x00aa6f},
+ {0x00aa70, 0x00aa70},
+ {0x00aa71, 0x00aa76},
+ {0x00aa7a, 0x00aa7a},
+ {0x00aa7b, 0x00aa7b},
+ {0x00aa7c, 0x00aa7c},
+ {0x00aa7d, 0x00aa7d},
+ {0x00aa7e, 0x00aaaf},
+ {0x00aab0, 0x00aab0},
+ {0x00aab1, 0x00aab1},
+ {0x00aab2, 0x00aab4},
+ {0x00aab5, 0x00aab6},
+ {0x00aab7, 0x00aab8},
+ {0x00aab9, 0x00aabd},
+ {0x00aabe, 0x00aabe},
+ {0x00aac0, 0x00aac0},
+ {0x00aac2, 0x00aac2},
+ {0x00aadb, 0x00aadc},
+ {0x00aadd, 0x00aadd},
+ {0x00aae0, 0x00aaea},
+ {0x00aaeb, 0x00aaeb},
+ {0x00aaec, 0x00aaed},
+ {0x00aaee, 0x00aaef},
+ {0x00aaf2, 0x00aaf2},
+ {0x00aaf3, 0x00aaf4},
+ {0x00aaf5, 0x00aaf5},
+ {0x00ab01, 0x00ab06},
+ {0x00ab09, 0x00ab0e},
+ {0x00ab11, 0x00ab16},
+ {0x00ab20, 0x00ab26},
+ {0x00ab28, 0x00ab2e},
+ {0x00ab30, 0x00ab5a},
+ {0x00ab5c, 0x00ab5f},
+ {0x00ab60, 0x00ab68},
+ {0x00ab69, 0x00ab69},
+ {0x00ab70, 0x00abbf},
+ {0x00abc0, 0x00abe2},
+ {0x00abe3, 0x00abe4},
+ {0x00abe5, 0x00abe5},
+ {0x00abe6, 0x00abe7},
+ {0x00abe8, 0x00abe8},
+ {0x00abe9, 0x00abea},
+ {0x00ac00, 0x00d7a3},
+ {0x00d7b0, 0x00d7c6},
+ {0x00d7cb, 0x00d7fb},
+ {0x00f900, 0x00fa6d},
+ {0x00fa70, 0x00fad9},
+ {0x00fb00, 0x00fb06},
+ {0x00fb13, 0x00fb17},
+ {0x00fb1d, 0x00fb1d},
+ {0x00fb1e, 0x00fb1e},
+ {0x00fb1f, 0x00fb28},
+ {0x00fb2a, 0x00fb36},
+ {0x00fb38, 0x00fb3c},
+ {0x00fb3e, 0x00fb3e},
+ {0x00fb40, 0x00fb41},
+ {0x00fb43, 0x00fb44},
+ {0x00fb46, 0x00fbb1},
+ {0x00fbd3, 0x00fd3d},
+ {0x00fd50, 0x00fd8f},
+ {0x00fd92, 0x00fdc7},
+ {0x00fdf0, 0x00fdfb},
+ {0x00fe70, 0x00fe74},
+ {0x00fe76, 0x00fefc},
+ {0x00ff21, 0x00ff3a},
+ {0x00ff41, 0x00ff5a},
+ {0x00ff66, 0x00ff6f},
+ {0x00ff70, 0x00ff70},
+ {0x00ff71, 0x00ff9d},
+ {0x00ff9e, 0x00ff9f},
+ {0x00ffa0, 0x00ffbe},
+ {0x00ffc2, 0x00ffc7},
+ {0x00ffca, 0x00ffcf},
+ {0x00ffd2, 0x00ffd7},
+ {0x00ffda, 0x00ffdc},
+ {0x010000, 0x01000b},
+ {0x01000d, 0x010026},
+ {0x010028, 0x01003a},
+ {0x01003c, 0x01003d},
+ {0x01003f, 0x01004d},
+ {0x010050, 0x01005d},
+ {0x010080, 0x0100fa},
+ {0x010140, 0x010174},
+ {0x010280, 0x01029c},
+ {0x0102a0, 0x0102d0},
+ {0x010300, 0x01031f},
+ {0x01032d, 0x010340},
+ {0x010341, 0x010341},
+ {0x010342, 0x010349},
+ {0x01034a, 0x01034a},
+ {0x010350, 0x010375},
+ {0x010376, 0x01037a},
+ {0x010380, 0x01039d},
+ {0x0103a0, 0x0103c3},
+ {0x0103c8, 0x0103cf},
+ {0x0103d1, 0x0103d5},
+ {0x010400, 0x01044f},
+ {0x010450, 0x01049d},
+ {0x0104b0, 0x0104d3},
+ {0x0104d8, 0x0104fb},
+ {0x010500, 0x010527},
+ {0x010530, 0x010563},
+ {0x010570, 0x01057a},
+ {0x01057c, 0x01058a},
+ {0x01058c, 0x010592},
+ {0x010594, 0x010595},
+ {0x010597, 0x0105a1},
+ {0x0105a3, 0x0105b1},
+ {0x0105b3, 0x0105b9},
+ {0x0105bb, 0x0105bc},
+ {0x010600, 0x010736},
+ {0x010740, 0x010755},
+ {0x010760, 0x010767},
+ {0x010780, 0x010785},
+ {0x010787, 0x0107b0},
+ {0x0107b2, 0x0107ba},
+ {0x010800, 0x010805},
+ {0x010808, 0x010808},
+ {0x01080a, 0x010835},
+ {0x010837, 0x010838},
+ {0x01083c, 0x01083c},
+ {0x01083f, 0x010855},
+ {0x010860, 0x010876},
+ {0x010880, 0x01089e},
+ {0x0108e0, 0x0108f2},
+ {0x0108f4, 0x0108f5},
+ {0x010900, 0x010915},
+ {0x010920, 0x010939},
+ {0x010980, 0x0109b7},
+ {0x0109be, 0x0109bf},
+ {0x010a00, 0x010a00},
+ {0x010a01, 0x010a03},
+ {0x010a05, 0x010a06},
+ {0x010a0c, 0x010a0f},
+ {0x010a10, 0x010a13},
+ {0x010a15, 0x010a17},
+ {0x010a19, 0x010a35},
+ {0x010a60, 0x010a7c},
+ {0x010a80, 0x010a9c},
+ {0x010ac0, 0x010ac7},
+ {0x010ac9, 0x010ae4},
+ {0x010b00, 0x010b35},
+ {0x010b40, 0x010b55},
+ {0x010b60, 0x010b72},
+ {0x010b80, 0x010b91},
+ {0x010c00, 0x010c48},
+ {0x010c80, 0x010cb2},
+ {0x010cc0, 0x010cf2},
+ {0x010d00, 0x010d23},
+ {0x010d24, 0x010d27},
+ {0x010e80, 0x010ea9},
+ {0x010eab, 0x010eac},
+ {0x010eb0, 0x010eb1},
+ {0x010f00, 0x010f1c},
+ {0x010f27, 0x010f27},
+ {0x010f30, 0x010f45},
+ {0x010f70, 0x010f81},
+ {0x010fb0, 0x010fc4},
+ {0x010fe0, 0x010ff6},
+ {0x011000, 0x011000},
+ {0x011001, 0x011001},
+ {0x011002, 0x011002},
+ {0x011003, 0x011037},
+ {0x011038, 0x011045},
+ {0x011071, 0x011072},
+ {0x011073, 0x011074},
+ {0x011075, 0x011075},
+ {0x011080, 0x011081},
+ {0x011082, 0x011082},
+ {0x011083, 0x0110af},
+ {0x0110b0, 0x0110b2},
+ {0x0110b3, 0x0110b6},
+ {0x0110b7, 0x0110b8},
+ {0x0110c2, 0x0110c2},
+ {0x0110d0, 0x0110e8},
+ {0x011100, 0x011102},
+ {0x011103, 0x011126},
+ {0x011127, 0x01112b},
+ {0x01112c, 0x01112c},
+ {0x01112d, 0x011132},
+ {0x011144, 0x011144},
+ {0x011145, 0x011146},
+ {0x011147, 0x011147},
+ {0x011150, 0x011172},
+ {0x011176, 0x011176},
+ {0x011180, 0x011181},
+ {0x011182, 0x011182},
+ {0x011183, 0x0111b2},
+ {0x0111b3, 0x0111b5},
+ {0x0111b6, 0x0111be},
+ {0x0111bf, 0x0111bf},
+ {0x0111c1, 0x0111c4},
+ {0x0111ce, 0x0111ce},
+ {0x0111cf, 0x0111cf},
+ {0x0111da, 0x0111da},
+ {0x0111dc, 0x0111dc},
+ {0x011200, 0x011211},
+ {0x011213, 0x01122b},
+ {0x01122c, 0x01122e},
+ {0x01122f, 0x011231},
+ {0x011232, 0x011233},
+ {0x011234, 0x011234},
+ {0x011237, 0x011237},
+ {0x01123e, 0x01123e},
+ {0x01123f, 0x011240},
+ {0x011241, 0x011241},
+ {0x011280, 0x011286},
+ {0x011288, 0x011288},
+ {0x01128a, 0x01128d},
+ {0x01128f, 0x01129d},
+ {0x01129f, 0x0112a8},
+ {0x0112b0, 0x0112de},
+ {0x0112df, 0x0112df},
+ {0x0112e0, 0x0112e2},
+ {0x0112e3, 0x0112e8},
+ {0x011300, 0x011301},
+ {0x011302, 0x011303},
+ {0x011305, 0x01130c},
+ {0x01130f, 0x011310},
+ {0x011313, 0x011328},
+ {0x01132a, 0x011330},
+ {0x011332, 0x011333},
+ {0x011335, 0x011339},
+ {0x01133d, 0x01133d},
+ {0x01133e, 0x01133f},
+ {0x011340, 0x011340},
+ {0x011341, 0x011344},
+ {0x011347, 0x011348},
+ {0x01134b, 0x01134c},
+ {0x011350, 0x011350},
+ {0x011357, 0x011357},
+ {0x01135d, 0x011361},
+ {0x011362, 0x011363},
+ {0x011400, 0x011434},
+ {0x011435, 0x011437},
+ {0x011438, 0x01143f},
+ {0x011440, 0x011441},
+ {0x011443, 0x011444},
+ {0x011445, 0x011445},
+ {0x011447, 0x01144a},
+ {0x01145f, 0x011461},
+ {0x011480, 0x0114af},
+ {0x0114b0, 0x0114b2},
+ {0x0114b3, 0x0114b8},
+ {0x0114b9, 0x0114b9},
+ {0x0114ba, 0x0114ba},
+ {0x0114bb, 0x0114be},
+ {0x0114bf, 0x0114c0},
+ {0x0114c1, 0x0114c1},
+ {0x0114c4, 0x0114c5},
+ {0x0114c7, 0x0114c7},
+ {0x011580, 0x0115ae},
+ {0x0115af, 0x0115b1},
+ {0x0115b2, 0x0115b5},
+ {0x0115b8, 0x0115bb},
+ {0x0115bc, 0x0115bd},
+ {0x0115be, 0x0115be},
+ {0x0115d8, 0x0115db},
+ {0x0115dc, 0x0115dd},
+ {0x011600, 0x01162f},
+ {0x011630, 0x011632},
+ {0x011633, 0x01163a},
+ {0x01163b, 0x01163c},
+ {0x01163d, 0x01163d},
+ {0x01163e, 0x01163e},
+ {0x011640, 0x011640},
+ {0x011644, 0x011644},
+ {0x011680, 0x0116aa},
+ {0x0116ab, 0x0116ab},
+ {0x0116ac, 0x0116ac},
+ {0x0116ad, 0x0116ad},
+ {0x0116ae, 0x0116af},
+ {0x0116b0, 0x0116b5},
+ {0x0116b8, 0x0116b8},
+ {0x011700, 0x01171a},
+ {0x01171d, 0x01171f},
+ {0x011720, 0x011721},
+ {0x011722, 0x011725},
+ {0x011726, 0x011726},
+ {0x011727, 0x01172a},
+ {0x011740, 0x011746},
+ {0x011800, 0x01182b},
+ {0x01182c, 0x01182e},
+ {0x01182f, 0x011837},
+ {0x011838, 0x011838},
+ {0x0118a0, 0x0118df},
+ {0x0118ff, 0x011906},
+ {0x011909, 0x011909},
+ {0x01190c, 0x011913},
+ {0x011915, 0x011916},
+ {0x011918, 0x01192f},
+ {0x011930, 0x011935},
+ {0x011937, 0x011938},
+ {0x01193b, 0x01193c},
+ {0x01193f, 0x01193f},
+ {0x011940, 0x011940},
+ {0x011941, 0x011941},
+ {0x011942, 0x011942},
+ {0x0119a0, 0x0119a7},
+ {0x0119aa, 0x0119d0},
+ {0x0119d1, 0x0119d3},
+ {0x0119d4, 0x0119d7},
+ {0x0119da, 0x0119db},
+ {0x0119dc, 0x0119df},
+ {0x0119e1, 0x0119e1},
+ {0x0119e3, 0x0119e3},
+ {0x0119e4, 0x0119e4},
+ {0x011a00, 0x011a00},
+ {0x011a01, 0x011a0a},
+ {0x011a0b, 0x011a32},
+ {0x011a35, 0x011a38},
+ {0x011a39, 0x011a39},
+ {0x011a3a, 0x011a3a},
+ {0x011a3b, 0x011a3e},
+ {0x011a50, 0x011a50},
+ {0x011a51, 0x011a56},
+ {0x011a57, 0x011a58},
+ {0x011a59, 0x011a5b},
+ {0x011a5c, 0x011a89},
+ {0x011a8a, 0x011a96},
+ {0x011a97, 0x011a97},
+ {0x011a9d, 0x011a9d},
+ {0x011ab0, 0x011af8},
+ {0x011c00, 0x011c08},
+ {0x011c0a, 0x011c2e},
+ {0x011c2f, 0x011c2f},
+ {0x011c30, 0x011c36},
+ {0x011c38, 0x011c3d},
+ {0x011c3e, 0x011c3e},
+ {0x011c40, 0x011c40},
+ {0x011c72, 0x011c8f},
+ {0x011c92, 0x011ca7},
+ {0x011ca9, 0x011ca9},
+ {0x011caa, 0x011cb0},
+ {0x011cb1, 0x011cb1},
+ {0x011cb2, 0x011cb3},
+ {0x011cb4, 0x011cb4},
+ {0x011cb5, 0x011cb6},
+ {0x011d00, 0x011d06},
+ {0x011d08, 0x011d09},
+ {0x011d0b, 0x011d30},
+ {0x011d31, 0x011d36},
+ {0x011d3a, 0x011d3a},
+ {0x011d3c, 0x011d3d},
+ {0x011d3f, 0x011d41},
+ {0x011d43, 0x011d43},
+ {0x011d46, 0x011d46},
+ {0x011d47, 0x011d47},
+ {0x011d60, 0x011d65},
+ {0x011d67, 0x011d68},
+ {0x011d6a, 0x011d89},
+ {0x011d8a, 0x011d8e},
+ {0x011d90, 0x011d91},
+ {0x011d93, 0x011d94},
+ {0x011d95, 0x011d95},
+ {0x011d96, 0x011d96},
+ {0x011d98, 0x011d98},
+ {0x011ee0, 0x011ef2},
+ {0x011ef3, 0x011ef4},
+ {0x011ef5, 0x011ef6},
+ {0x011f00, 0x011f01},
+ {0x011f02, 0x011f02},
+ {0x011f03, 0x011f03},
+ {0x011f04, 0x011f10},
+ {0x011f12, 0x011f33},
+ {0x011f34, 0x011f35},
+ {0x011f36, 0x011f3a},
+ {0x011f3e, 0x011f3f},
+ {0x011f40, 0x011f40},
+ {0x011fb0, 0x011fb0},
+ {0x012000, 0x012399},
+ {0x012400, 0x01246e},
+ {0x012480, 0x012543},
+ {0x012f90, 0x012ff0},
+ {0x013000, 0x01342f},
+ {0x013441, 0x013446},
+ {0x014400, 0x014646},
+ {0x016800, 0x016a38},
+ {0x016a40, 0x016a5e},
+ {0x016a70, 0x016abe},
+ {0x016ad0, 0x016aed},
+ {0x016b00, 0x016b2f},
+ {0x016b40, 0x016b43},
+ {0x016b63, 0x016b77},
+ {0x016b7d, 0x016b8f},
+ {0x016e40, 0x016e7f},
+ {0x016f00, 0x016f4a},
+ {0x016f4f, 0x016f4f},
+ {0x016f50, 0x016f50},
+ {0x016f51, 0x016f87},
+ {0x016f8f, 0x016f92},
+ {0x016f93, 0x016f9f},
+ {0x016fe0, 0x016fe1},
+ {0x016fe3, 0x016fe3},
+ {0x016ff0, 0x016ff1},
+ {0x017000, 0x0187f7},
+ {0x018800, 0x018cd5},
+ {0x018d00, 0x018d08},
+ {0x01aff0, 0x01aff3},
+ {0x01aff5, 0x01affb},
+ {0x01affd, 0x01affe},
+ {0x01b000, 0x01b122},
+ {0x01b132, 0x01b132},
+ {0x01b150, 0x01b152},
+ {0x01b155, 0x01b155},
+ {0x01b164, 0x01b167},
+ {0x01b170, 0x01b2fb},
+ {0x01bc00, 0x01bc6a},
+ {0x01bc70, 0x01bc7c},
+ {0x01bc80, 0x01bc88},
+ {0x01bc90, 0x01bc99},
+ {0x01bc9e, 0x01bc9e},
+ {0x01d400, 0x01d454},
+ {0x01d456, 0x01d49c},
+ {0x01d49e, 0x01d49f},
+ {0x01d4a2, 0x01d4a2},
+ {0x01d4a5, 0x01d4a6},
+ {0x01d4a9, 0x01d4ac},
+ {0x01d4ae, 0x01d4b9},
+ {0x01d4bb, 0x01d4bb},
+ {0x01d4bd, 0x01d4c3},
+ {0x01d4c5, 0x01d505},
+ {0x01d507, 0x01d50a},
+ {0x01d50d, 0x01d514},
+ {0x01d516, 0x01d51c},
+ {0x01d51e, 0x01d539},
+ {0x01d53b, 0x01d53e},
+ {0x01d540, 0x01d544},
+ {0x01d546, 0x01d546},
+ {0x01d54a, 0x01d550},
+ {0x01d552, 0x01d6a5},
+ {0x01d6a8, 0x01d6c0},
+ {0x01d6c2, 0x01d6da},
+ {0x01d6dc, 0x01d6fa},
+ {0x01d6fc, 0x01d714},
+ {0x01d716, 0x01d734},
+ {0x01d736, 0x01d74e},
+ {0x01d750, 0x01d76e},
+ {0x01d770, 0x01d788},
+ {0x01d78a, 0x01d7a8},
+ {0x01d7aa, 0x01d7c2},
+ {0x01d7c4, 0x01d7cb},
+ {0x01df00, 0x01df09},
+ {0x01df0a, 0x01df0a},
+ {0x01df0b, 0x01df1e},
+ {0x01df25, 0x01df2a},
+ {0x01e000, 0x01e006},
+ {0x01e008, 0x01e018},
+ {0x01e01b, 0x01e021},
+ {0x01e023, 0x01e024},
+ {0x01e026, 0x01e02a},
+ {0x01e030, 0x01e06d},
+ {0x01e08f, 0x01e08f},
+ {0x01e100, 0x01e12c},
+ {0x01e137, 0x01e13d},
+ {0x01e14e, 0x01e14e},
+ {0x01e290, 0x01e2ad},
+ {0x01e2c0, 0x01e2eb},
+ {0x01e4d0, 0x01e4ea},
+ {0x01e4eb, 0x01e4eb},
+ {0x01e7e0, 0x01e7e6},
+ {0x01e7e8, 0x01e7eb},
+ {0x01e7ed, 0x01e7ee},
+ {0x01e7f0, 0x01e7fe},
+ {0x01e800, 0x01e8c4},
+ {0x01e900, 0x01e943},
+ {0x01e947, 0x01e947},
+ {0x01e94b, 0x01e94b},
+ {0x01ee00, 0x01ee03},
+ {0x01ee05, 0x01ee1f},
+ {0x01ee21, 0x01ee22},
+ {0x01ee24, 0x01ee24},
+ {0x01ee27, 0x01ee27},
+ {0x01ee29, 0x01ee32},
+ {0x01ee34, 0x01ee37},
+ {0x01ee39, 0x01ee39},
+ {0x01ee3b, 0x01ee3b},
+ {0x01ee42, 0x01ee42},
+ {0x01ee47, 0x01ee47},
+ {0x01ee49, 0x01ee49},
+ {0x01ee4b, 0x01ee4b},
+ {0x01ee4d, 0x01ee4f},
+ {0x01ee51, 0x01ee52},
+ {0x01ee54, 0x01ee54},
+ {0x01ee57, 0x01ee57},
+ {0x01ee59, 0x01ee59},
+ {0x01ee5b, 0x01ee5b},
+ {0x01ee5d, 0x01ee5d},
+ {0x01ee5f, 0x01ee5f},
+ {0x01ee61, 0x01ee62},
+ {0x01ee64, 0x01ee64},
+ {0x01ee67, 0x01ee6a},
+ {0x01ee6c, 0x01ee72},
+ {0x01ee74, 0x01ee77},
+ {0x01ee79, 0x01ee7c},
+ {0x01ee7e, 0x01ee7e},
+ {0x01ee80, 0x01ee89},
+ {0x01ee8b, 0x01ee9b},
+ {0x01eea1, 0x01eea3},
+ {0x01eea5, 0x01eea9},
+ {0x01eeab, 0x01eebb},
+ {0x01f130, 0x01f149},
+ {0x01f150, 0x01f169},
+ {0x01f170, 0x01f189},
+ {0x020000, 0x02a6df},
+ {0x02a700, 0x02b739},
+ {0x02b740, 0x02b81d},
+ {0x02b820, 0x02cea1},
+ {0x02ceb0, 0x02ebe0},
+ {0x02ebf0, 0x02ee5d},
+ {0x02f800, 0x02fa1d},
+ {0x030000, 0x03134a},
+ {0x031350, 0x0323af}
+};
+
+/* table of Unicode codepoint ranges of Lowercase characters */
+static const pg_unicode_range unicode_lowercase[686] =
+{
+ {0x000061, 0x00007a},
+ {0x0000aa, 0x0000aa},
+ {0x0000b5, 0x0000b5},
+ {0x0000ba, 0x0000ba},
+ {0x0000df, 0x0000f6},
+ {0x0000f8, 0x0000ff},
+ {0x000101, 0x000101},
+ {0x000103, 0x000103},
+ {0x000105, 0x000105},
+ {0x000107, 0x000107},
+ {0x000109, 0x000109},
+ {0x00010b, 0x00010b},
+ {0x00010d, 0x00010d},
+ {0x00010f, 0x00010f},
+ {0x000111, 0x000111},
+ {0x000113, 0x000113},
+ {0x000115, 0x000115},
+ {0x000117, 0x000117},
+ {0x000119, 0x000119},
+ {0x00011b, 0x00011b},
+ {0x00011d, 0x00011d},
+ {0x00011f, 0x00011f},
+ {0x000121, 0x000121},
+ {0x000123, 0x000123},
+ {0x000125, 0x000125},
+ {0x000127, 0x000127},
+ {0x000129, 0x000129},
+ {0x00012b, 0x00012b},
+ {0x00012d, 0x00012d},
+ {0x00012f, 0x00012f},
+ {0x000131, 0x000131},
+ {0x000133, 0x000133},
+ {0x000135, 0x000135},
+ {0x000137, 0x000138},
+ {0x00013a, 0x00013a},
+ {0x00013c, 0x00013c},
+ {0x00013e, 0x00013e},
+ {0x000140, 0x000140},
+ {0x000142, 0x000142},
+ {0x000144, 0x000144},
+ {0x000146, 0x000146},
+ {0x000148, 0x000149},
+ {0x00014b, 0x00014b},
+ {0x00014d, 0x00014d},
+ {0x00014f, 0x00014f},
+ {0x000151, 0x000151},
+ {0x000153, 0x000153},
+ {0x000155, 0x000155},
+ {0x000157, 0x000157},
+ {0x000159, 0x000159},
+ {0x00015b, 0x00015b},
+ {0x00015d, 0x00015d},
+ {0x00015f, 0x00015f},
+ {0x000161, 0x000161},
+ {0x000163, 0x000163},
+ {0x000165, 0x000165},
+ {0x000167, 0x000167},
+ {0x000169, 0x000169},
+ {0x00016b, 0x00016b},
+ {0x00016d, 0x00016d},
+ {0x00016f, 0x00016f},
+ {0x000171, 0x000171},
+ {0x000173, 0x000173},
+ {0x000175, 0x000175},
+ {0x000177, 0x000177},
+ {0x00017a, 0x00017a},
+ {0x00017c, 0x00017c},
+ {0x00017e, 0x000180},
+ {0x000183, 0x000183},
+ {0x000185, 0x000185},
+ {0x000188, 0x000188},
+ {0x00018c, 0x00018d},
+ {0x000192, 0x000192},
+ {0x000195, 0x000195},
+ {0x000199, 0x00019b},
+ {0x00019e, 0x00019e},
+ {0x0001a1, 0x0001a1},
+ {0x0001a3, 0x0001a3},
+ {0x0001a5, 0x0001a5},
+ {0x0001a8, 0x0001a8},
+ {0x0001aa, 0x0001ab},
+ {0x0001ad, 0x0001ad},
+ {0x0001b0, 0x0001b0},
+ {0x0001b4, 0x0001b4},
+ {0x0001b6, 0x0001b6},
+ {0x0001b9, 0x0001ba},
+ {0x0001bd, 0x0001bf},
+ {0x0001c6, 0x0001c6},
+ {0x0001c9, 0x0001c9},
+ {0x0001cc, 0x0001cc},
+ {0x0001ce, 0x0001ce},
+ {0x0001d0, 0x0001d0},
+ {0x0001d2, 0x0001d2},
+ {0x0001d4, 0x0001d4},
+ {0x0001d6, 0x0001d6},
+ {0x0001d8, 0x0001d8},
+ {0x0001da, 0x0001da},
+ {0x0001dc, 0x0001dd},
+ {0x0001df, 0x0001df},
+ {0x0001e1, 0x0001e1},
+ {0x0001e3, 0x0001e3},
+ {0x0001e5, 0x0001e5},
+ {0x0001e7, 0x0001e7},
+ {0x0001e9, 0x0001e9},
+ {0x0001eb, 0x0001eb},
+ {0x0001ed, 0x0001ed},
+ {0x0001ef, 0x0001f0},
+ {0x0001f3, 0x0001f3},
+ {0x0001f5, 0x0001f5},
+ {0x0001f9, 0x0001f9},
+ {0x0001fb, 0x0001fb},
+ {0x0001fd, 0x0001fd},
+ {0x0001ff, 0x0001ff},
+ {0x000201, 0x000201},
+ {0x000203, 0x000203},
+ {0x000205, 0x000205},
+ {0x000207, 0x000207},
+ {0x000209, 0x000209},
+ {0x00020b, 0x00020b},
+ {0x00020d, 0x00020d},
+ {0x00020f, 0x00020f},
+ {0x000211, 0x000211},
+ {0x000213, 0x000213},
+ {0x000215, 0x000215},
+ {0x000217, 0x000217},
+ {0x000219, 0x000219},
+ {0x00021b, 0x00021b},
+ {0x00021d, 0x00021d},
+ {0x00021f, 0x00021f},
+ {0x000221, 0x000221},
+ {0x000223, 0x000223},
+ {0x000225, 0x000225},
+ {0x000227, 0x000227},
+ {0x000229, 0x000229},
+ {0x00022b, 0x00022b},
+ {0x00022d, 0x00022d},
+ {0x00022f, 0x00022f},
+ {0x000231, 0x000231},
+ {0x000233, 0x000239},
+ {0x00023c, 0x00023c},
+ {0x00023f, 0x000240},
+ {0x000242, 0x000242},
+ {0x000247, 0x000247},
+ {0x000249, 0x000249},
+ {0x00024b, 0x00024b},
+ {0x00024d, 0x00024d},
+ {0x00024f, 0x000293},
+ {0x000295, 0x0002af},
+ {0x0002b0, 0x0002b8},
+ {0x0002c0, 0x0002c1},
+ {0x0002e0, 0x0002e4},
+ {0x000345, 0x000345},
+ {0x000371, 0x000371},
+ {0x000373, 0x000373},
+ {0x000377, 0x000377},
+ {0x00037a, 0x00037a},
+ {0x00037b, 0x00037d},
+ {0x000390, 0x000390},
+ {0x0003ac, 0x0003ce},
+ {0x0003d0, 0x0003d1},
+ {0x0003d5, 0x0003d7},
+ {0x0003d9, 0x0003d9},
+ {0x0003db, 0x0003db},
+ {0x0003dd, 0x0003dd},
+ {0x0003df, 0x0003df},
+ {0x0003e1, 0x0003e1},
+ {0x0003e3, 0x0003e3},
+ {0x0003e5, 0x0003e5},
+ {0x0003e7, 0x0003e7},
+ {0x0003e9, 0x0003e9},
+ {0x0003eb, 0x0003eb},
+ {0x0003ed, 0x0003ed},
+ {0x0003ef, 0x0003f3},
+ {0x0003f5, 0x0003f5},
+ {0x0003f8, 0x0003f8},
+ {0x0003fb, 0x0003fc},
+ {0x000430, 0x00045f},
+ {0x000461, 0x000461},
+ {0x000463, 0x000463},
+ {0x000465, 0x000465},
+ {0x000467, 0x000467},
+ {0x000469, 0x000469},
+ {0x00046b, 0x00046b},
+ {0x00046d, 0x00046d},
+ {0x00046f, 0x00046f},
+ {0x000471, 0x000471},
+ {0x000473, 0x000473},
+ {0x000475, 0x000475},
+ {0x000477, 0x000477},
+ {0x000479, 0x000479},
+ {0x00047b, 0x00047b},
+ {0x00047d, 0x00047d},
+ {0x00047f, 0x00047f},
+ {0x000481, 0x000481},
+ {0x00048b, 0x00048b},
+ {0x00048d, 0x00048d},
+ {0x00048f, 0x00048f},
+ {0x000491, 0x000491},
+ {0x000493, 0x000493},
+ {0x000495, 0x000495},
+ {0x000497, 0x000497},
+ {0x000499, 0x000499},
+ {0x00049b, 0x00049b},
+ {0x00049d, 0x00049d},
+ {0x00049f, 0x00049f},
+ {0x0004a1, 0x0004a1},
+ {0x0004a3, 0x0004a3},
+ {0x0004a5, 0x0004a5},
+ {0x0004a7, 0x0004a7},
+ {0x0004a9, 0x0004a9},
+ {0x0004ab, 0x0004ab},
+ {0x0004ad, 0x0004ad},
+ {0x0004af, 0x0004af},
+ {0x0004b1, 0x0004b1},
+ {0x0004b3, 0x0004b3},
+ {0x0004b5, 0x0004b5},
+ {0x0004b7, 0x0004b7},
+ {0x0004b9, 0x0004b9},
+ {0x0004bb, 0x0004bb},
+ {0x0004bd, 0x0004bd},
+ {0x0004bf, 0x0004bf},
+ {0x0004c2, 0x0004c2},
+ {0x0004c4, 0x0004c4},
+ {0x0004c6, 0x0004c6},
+ {0x0004c8, 0x0004c8},
+ {0x0004ca, 0x0004ca},
+ {0x0004cc, 0x0004cc},
+ {0x0004ce, 0x0004cf},
+ {0x0004d1, 0x0004d1},
+ {0x0004d3, 0x0004d3},
+ {0x0004d5, 0x0004d5},
+ {0x0004d7, 0x0004d7},
+ {0x0004d9, 0x0004d9},
+ {0x0004db, 0x0004db},
+ {0x0004dd, 0x0004dd},
+ {0x0004df, 0x0004df},
+ {0x0004e1, 0x0004e1},
+ {0x0004e3, 0x0004e3},
+ {0x0004e5, 0x0004e5},
+ {0x0004e7, 0x0004e7},
+ {0x0004e9, 0x0004e9},
+ {0x0004eb, 0x0004eb},
+ {0x0004ed, 0x0004ed},
+ {0x0004ef, 0x0004ef},
+ {0x0004f1, 0x0004f1},
+ {0x0004f3, 0x0004f3},
+ {0x0004f5, 0x0004f5},
+ {0x0004f7, 0x0004f7},
+ {0x0004f9, 0x0004f9},
+ {0x0004fb, 0x0004fb},
+ {0x0004fd, 0x0004fd},
+ {0x0004ff, 0x0004ff},
+ {0x000501, 0x000501},
+ {0x000503, 0x000503},
+ {0x000505, 0x000505},
+ {0x000507, 0x000507},
+ {0x000509, 0x000509},
+ {0x00050b, 0x00050b},
+ {0x00050d, 0x00050d},
+ {0x00050f, 0x00050f},
+ {0x000511, 0x000511},
+ {0x000513, 0x000513},
+ {0x000515, 0x000515},
+ {0x000517, 0x000517},
+ {0x000519, 0x000519},
+ {0x00051b, 0x00051b},
+ {0x00051d, 0x00051d},
+ {0x00051f, 0x00051f},
+ {0x000521, 0x000521},
+ {0x000523, 0x000523},
+ {0x000525, 0x000525},
+ {0x000527, 0x000527},
+ {0x000529, 0x000529},
+ {0x00052b, 0x00052b},
+ {0x00052d, 0x00052d},
+ {0x00052f, 0x00052f},
+ {0x000560, 0x000588},
+ {0x0010d0, 0x0010fa},
+ {0x0010fc, 0x0010fc},
+ {0x0010fd, 0x0010ff},
+ {0x0013f8, 0x0013fd},
+ {0x001c80, 0x001c88},
+ {0x001d00, 0x001d2b},
+ {0x001d2c, 0x001d6a},
+ {0x001d6b, 0x001d77},
+ {0x001d78, 0x001d78},
+ {0x001d79, 0x001d9a},
+ {0x001d9b, 0x001dbf},
+ {0x001e01, 0x001e01},
+ {0x001e03, 0x001e03},
+ {0x001e05, 0x001e05},
+ {0x001e07, 0x001e07},
+ {0x001e09, 0x001e09},
+ {0x001e0b, 0x001e0b},
+ {0x001e0d, 0x001e0d},
+ {0x001e0f, 0x001e0f},
+ {0x001e11, 0x001e11},
+ {0x001e13, 0x001e13},
+ {0x001e15, 0x001e15},
+ {0x001e17, 0x001e17},
+ {0x001e19, 0x001e19},
+ {0x001e1b, 0x001e1b},
+ {0x001e1d, 0x001e1d},
+ {0x001e1f, 0x001e1f},
+ {0x001e21, 0x001e21},
+ {0x001e23, 0x001e23},
+ {0x001e25, 0x001e25},
+ {0x001e27, 0x001e27},
+ {0x001e29, 0x001e29},
+ {0x001e2b, 0x001e2b},
+ {0x001e2d, 0x001e2d},
+ {0x001e2f, 0x001e2f},
+ {0x001e31, 0x001e31},
+ {0x001e33, 0x001e33},
+ {0x001e35, 0x001e35},
+ {0x001e37, 0x001e37},
+ {0x001e39, 0x001e39},
+ {0x001e3b, 0x001e3b},
+ {0x001e3d, 0x001e3d},
+ {0x001e3f, 0x001e3f},
+ {0x001e41, 0x001e41},
+ {0x001e43, 0x001e43},
+ {0x001e45, 0x001e45},
+ {0x001e47, 0x001e47},
+ {0x001e49, 0x001e49},
+ {0x001e4b, 0x001e4b},
+ {0x001e4d, 0x001e4d},
+ {0x001e4f, 0x001e4f},
+ {0x001e51, 0x001e51},
+ {0x001e53, 0x001e53},
+ {0x001e55, 0x001e55},
+ {0x001e57, 0x001e57},
+ {0x001e59, 0x001e59},
+ {0x001e5b, 0x001e5b},
+ {0x001e5d, 0x001e5d},
+ {0x001e5f, 0x001e5f},
+ {0x001e61, 0x001e61},
+ {0x001e63, 0x001e63},
+ {0x001e65, 0x001e65},
+ {0x001e67, 0x001e67},
+ {0x001e69, 0x001e69},
+ {0x001e6b, 0x001e6b},
+ {0x001e6d, 0x001e6d},
+ {0x001e6f, 0x001e6f},
+ {0x001e71, 0x001e71},
+ {0x001e73, 0x001e73},
+ {0x001e75, 0x001e75},
+ {0x001e77, 0x001e77},
+ {0x001e79, 0x001e79},
+ {0x001e7b, 0x001e7b},
+ {0x001e7d, 0x001e7d},
+ {0x001e7f, 0x001e7f},
+ {0x001e81, 0x001e81},
+ {0x001e83, 0x001e83},
+ {0x001e85, 0x001e85},
+ {0x001e87, 0x001e87},
+ {0x001e89, 0x001e89},
+ {0x001e8b, 0x001e8b},
+ {0x001e8d, 0x001e8d},
+ {0x001e8f, 0x001e8f},
+ {0x001e91, 0x001e91},
+ {0x001e93, 0x001e93},
+ {0x001e95, 0x001e9d},
+ {0x001e9f, 0x001e9f},
+ {0x001ea1, 0x001ea1},
+ {0x001ea3, 0x001ea3},
+ {0x001ea5, 0x001ea5},
+ {0x001ea7, 0x001ea7},
+ {0x001ea9, 0x001ea9},
+ {0x001eab, 0x001eab},
+ {0x001ead, 0x001ead},
+ {0x001eaf, 0x001eaf},
+ {0x001eb1, 0x001eb1},
+ {0x001eb3, 0x001eb3},
+ {0x001eb5, 0x001eb5},
+ {0x001eb7, 0x001eb7},
+ {0x001eb9, 0x001eb9},
+ {0x001ebb, 0x001ebb},
+ {0x001ebd, 0x001ebd},
+ {0x001ebf, 0x001ebf},
+ {0x001ec1, 0x001ec1},
+ {0x001ec3, 0x001ec3},
+ {0x001ec5, 0x001ec5},
+ {0x001ec7, 0x001ec7},
+ {0x001ec9, 0x001ec9},
+ {0x001ecb, 0x001ecb},
+ {0x001ecd, 0x001ecd},
+ {0x001ecf, 0x001ecf},
+ {0x001ed1, 0x001ed1},
+ {0x001ed3, 0x001ed3},
+ {0x001ed5, 0x001ed5},
+ {0x001ed7, 0x001ed7},
+ {0x001ed9, 0x001ed9},
+ {0x001edb, 0x001edb},
+ {0x001edd, 0x001edd},
+ {0x001edf, 0x001edf},
+ {0x001ee1, 0x001ee1},
+ {0x001ee3, 0x001ee3},
+ {0x001ee5, 0x001ee5},
+ {0x001ee7, 0x001ee7},
+ {0x001ee9, 0x001ee9},
+ {0x001eeb, 0x001eeb},
+ {0x001eed, 0x001eed},
+ {0x001eef, 0x001eef},
+ {0x001ef1, 0x001ef1},
+ {0x001ef3, 0x001ef3},
+ {0x001ef5, 0x001ef5},
+ {0x001ef7, 0x001ef7},
+ {0x001ef9, 0x001ef9},
+ {0x001efb, 0x001efb},
+ {0x001efd, 0x001efd},
+ {0x001eff, 0x001f07},
+ {0x001f10, 0x001f15},
+ {0x001f20, 0x001f27},
+ {0x001f30, 0x001f37},
+ {0x001f40, 0x001f45},
+ {0x001f50, 0x001f57},
+ {0x001f60, 0x001f67},
+ {0x001f70, 0x001f7d},
+ {0x001f80, 0x001f87},
+ {0x001f90, 0x001f97},
+ {0x001fa0, 0x001fa7},
+ {0x001fb0, 0x001fb4},
+ {0x001fb6, 0x001fb7},
+ {0x001fbe, 0x001fbe},
+ {0x001fc2, 0x001fc4},
+ {0x001fc6, 0x001fc7},
+ {0x001fd0, 0x001fd3},
+ {0x001fd6, 0x001fd7},
+ {0x001fe0, 0x001fe7},
+ {0x001ff2, 0x001ff4},
+ {0x001ff6, 0x001ff7},
+ {0x002071, 0x002071},
+ {0x00207f, 0x00207f},
+ {0x002090, 0x00209c},
+ {0x00210a, 0x00210a},
+ {0x00210e, 0x00210f},
+ {0x002113, 0x002113},
+ {0x00212f, 0x00212f},
+ {0x002134, 0x002134},
+ {0x002139, 0x002139},
+ {0x00213c, 0x00213d},
+ {0x002146, 0x002149},
+ {0x00214e, 0x00214e},
+ {0x002170, 0x00217f},
+ {0x002184, 0x002184},
+ {0x0024d0, 0x0024e9},
+ {0x002c30, 0x002c5f},
+ {0x002c61, 0x002c61},
+ {0x002c65, 0x002c66},
+ {0x002c68, 0x002c68},
+ {0x002c6a, 0x002c6a},
+ {0x002c6c, 0x002c6c},
+ {0x002c71, 0x002c71},
+ {0x002c73, 0x002c74},
+ {0x002c76, 0x002c7b},
+ {0x002c7c, 0x002c7d},
+ {0x002c81, 0x002c81},
+ {0x002c83, 0x002c83},
+ {0x002c85, 0x002c85},
+ {0x002c87, 0x002c87},
+ {0x002c89, 0x002c89},
+ {0x002c8b, 0x002c8b},
+ {0x002c8d, 0x002c8d},
+ {0x002c8f, 0x002c8f},
+ {0x002c91, 0x002c91},
+ {0x002c93, 0x002c93},
+ {0x002c95, 0x002c95},
+ {0x002c97, 0x002c97},
+ {0x002c99, 0x002c99},
+ {0x002c9b, 0x002c9b},
+ {0x002c9d, 0x002c9d},
+ {0x002c9f, 0x002c9f},
+ {0x002ca1, 0x002ca1},
+ {0x002ca3, 0x002ca3},
+ {0x002ca5, 0x002ca5},
+ {0x002ca7, 0x002ca7},
+ {0x002ca9, 0x002ca9},
+ {0x002cab, 0x002cab},
+ {0x002cad, 0x002cad},
+ {0x002caf, 0x002caf},
+ {0x002cb1, 0x002cb1},
+ {0x002cb3, 0x002cb3},
+ {0x002cb5, 0x002cb5},
+ {0x002cb7, 0x002cb7},
+ {0x002cb9, 0x002cb9},
+ {0x002cbb, 0x002cbb},
+ {0x002cbd, 0x002cbd},
+ {0x002cbf, 0x002cbf},
+ {0x002cc1, 0x002cc1},
+ {0x002cc3, 0x002cc3},
+ {0x002cc5, 0x002cc5},
+ {0x002cc7, 0x002cc7},
+ {0x002cc9, 0x002cc9},
+ {0x002ccb, 0x002ccb},
+ {0x002ccd, 0x002ccd},
+ {0x002ccf, 0x002ccf},
+ {0x002cd1, 0x002cd1},
+ {0x002cd3, 0x002cd3},
+ {0x002cd5, 0x002cd5},
+ {0x002cd7, 0x002cd7},
+ {0x002cd9, 0x002cd9},
+ {0x002cdb, 0x002cdb},
+ {0x002cdd, 0x002cdd},
+ {0x002cdf, 0x002cdf},
+ {0x002ce1, 0x002ce1},
+ {0x002ce3, 0x002ce4},
+ {0x002cec, 0x002cec},
+ {0x002cee, 0x002cee},
+ {0x002cf3, 0x002cf3},
+ {0x002d00, 0x002d25},
+ {0x002d27, 0x002d27},
+ {0x002d2d, 0x002d2d},
+ {0x00a641, 0x00a641},
+ {0x00a643, 0x00a643},
+ {0x00a645, 0x00a645},
+ {0x00a647, 0x00a647},
+ {0x00a649, 0x00a649},
+ {0x00a64b, 0x00a64b},
+ {0x00a64d, 0x00a64d},
+ {0x00a64f, 0x00a64f},
+ {0x00a651, 0x00a651},
+ {0x00a653, 0x00a653},
+ {0x00a655, 0x00a655},
+ {0x00a657, 0x00a657},
+ {0x00a659, 0x00a659},
+ {0x00a65b, 0x00a65b},
+ {0x00a65d, 0x00a65d},
+ {0x00a65f, 0x00a65f},
+ {0x00a661, 0x00a661},
+ {0x00a663, 0x00a663},
+ {0x00a665, 0x00a665},
+ {0x00a667, 0x00a667},
+ {0x00a669, 0x00a669},
+ {0x00a66b, 0x00a66b},
+ {0x00a66d, 0x00a66d},
+ {0x00a681, 0x00a681},
+ {0x00a683, 0x00a683},
+ {0x00a685, 0x00a685},
+ {0x00a687, 0x00a687},
+ {0x00a689, 0x00a689},
+ {0x00a68b, 0x00a68b},
+ {0x00a68d, 0x00a68d},
+ {0x00a68f, 0x00a68f},
+ {0x00a691, 0x00a691},
+ {0x00a693, 0x00a693},
+ {0x00a695, 0x00a695},
+ {0x00a697, 0x00a697},
+ {0x00a699, 0x00a699},
+ {0x00a69b, 0x00a69b},
+ {0x00a69c, 0x00a69d},
+ {0x00a723, 0x00a723},
+ {0x00a725, 0x00a725},
+ {0x00a727, 0x00a727},
+ {0x00a729, 0x00a729},
+ {0x00a72b, 0x00a72b},
+ {0x00a72d, 0x00a72d},
+ {0x00a72f, 0x00a731},
+ {0x00a733, 0x00a733},
+ {0x00a735, 0x00a735},
+ {0x00a737, 0x00a737},
+ {0x00a739, 0x00a739},
+ {0x00a73b, 0x00a73b},
+ {0x00a73d, 0x00a73d},
+ {0x00a73f, 0x00a73f},
+ {0x00a741, 0x00a741},
+ {0x00a743, 0x00a743},
+ {0x00a745, 0x00a745},
+ {0x00a747, 0x00a747},
+ {0x00a749, 0x00a749},
+ {0x00a74b, 0x00a74b},
+ {0x00a74d, 0x00a74d},
+ {0x00a74f, 0x00a74f},
+ {0x00a751, 0x00a751},
+ {0x00a753, 0x00a753},
+ {0x00a755, 0x00a755},
+ {0x00a757, 0x00a757},
+ {0x00a759, 0x00a759},
+ {0x00a75b, 0x00a75b},
+ {0x00a75d, 0x00a75d},
+ {0x00a75f, 0x00a75f},
+ {0x00a761, 0x00a761},
+ {0x00a763, 0x00a763},
+ {0x00a765, 0x00a765},
+ {0x00a767, 0x00a767},
+ {0x00a769, 0x00a769},
+ {0x00a76b, 0x00a76b},
+ {0x00a76d, 0x00a76d},
+ {0x00a76f, 0x00a76f},
+ {0x00a770, 0x00a770},
+ {0x00a771, 0x00a778},
+ {0x00a77a, 0x00a77a},
+ {0x00a77c, 0x00a77c},
+ {0x00a77f, 0x00a77f},
+ {0x00a781, 0x00a781},
+ {0x00a783, 0x00a783},
+ {0x00a785, 0x00a785},
+ {0x00a787, 0x00a787},
+ {0x00a78c, 0x00a78c},
+ {0x00a78e, 0x00a78e},
+ {0x00a791, 0x00a791},
+ {0x00a793, 0x00a795},
+ {0x00a797, 0x00a797},
+ {0x00a799, 0x00a799},
+ {0x00a79b, 0x00a79b},
+ {0x00a79d, 0x00a79d},
+ {0x00a79f, 0x00a79f},
+ {0x00a7a1, 0x00a7a1},
+ {0x00a7a3, 0x00a7a3},
+ {0x00a7a5, 0x00a7a5},
+ {0x00a7a7, 0x00a7a7},
+ {0x00a7a9, 0x00a7a9},
+ {0x00a7af, 0x00a7af},
+ {0x00a7b5, 0x00a7b5},
+ {0x00a7b7, 0x00a7b7},
+ {0x00a7b9, 0x00a7b9},
+ {0x00a7bb, 0x00a7bb},
+ {0x00a7bd, 0x00a7bd},
+ {0x00a7bf, 0x00a7bf},
+ {0x00a7c1, 0x00a7c1},
+ {0x00a7c3, 0x00a7c3},
+ {0x00a7c8, 0x00a7c8},
+ {0x00a7ca, 0x00a7ca},
+ {0x00a7d1, 0x00a7d1},
+ {0x00a7d3, 0x00a7d3},
+ {0x00a7d5, 0x00a7d5},
+ {0x00a7d7, 0x00a7d7},
+ {0x00a7d9, 0x00a7d9},
+ {0x00a7f2, 0x00a7f4},
+ {0x00a7f6, 0x00a7f6},
+ {0x00a7f8, 0x00a7f9},
+ {0x00a7fa, 0x00a7fa},
+ {0x00ab30, 0x00ab5a},
+ {0x00ab5c, 0x00ab5f},
+ {0x00ab60, 0x00ab68},
+ {0x00ab69, 0x00ab69},
+ {0x00ab70, 0x00abbf},
+ {0x00fb00, 0x00fb06},
+ {0x00fb13, 0x00fb17},
+ {0x00ff41, 0x00ff5a},
+ {0x010428, 0x01044f},
+ {0x0104d8, 0x0104fb},
+ {0x010597, 0x0105a1},
+ {0x0105a3, 0x0105b1},
+ {0x0105b3, 0x0105b9},
+ {0x0105bb, 0x0105bc},
+ {0x010780, 0x010780},
+ {0x010783, 0x010785},
+ {0x010787, 0x0107b0},
+ {0x0107b2, 0x0107ba},
+ {0x010cc0, 0x010cf2},
+ {0x0118c0, 0x0118df},
+ {0x016e60, 0x016e7f},
+ {0x01d41a, 0x01d433},
+ {0x01d44e, 0x01d454},
+ {0x01d456, 0x01d467},
+ {0x01d482, 0x01d49b},
+ {0x01d4b6, 0x01d4b9},
+ {0x01d4bb, 0x01d4bb},
+ {0x01d4bd, 0x01d4c3},
+ {0x01d4c5, 0x01d4cf},
+ {0x01d4ea, 0x01d503},
+ {0x01d51e, 0x01d537},
+ {0x01d552, 0x01d56b},
+ {0x01d586, 0x01d59f},
+ {0x01d5ba, 0x01d5d3},
+ {0x01d5ee, 0x01d607},
+ {0x01d622, 0x01d63b},
+ {0x01d656, 0x01d66f},
+ {0x01d68a, 0x01d6a5},
+ {0x01d6c2, 0x01d6da},
+ {0x01d6dc, 0x01d6e1},
+ {0x01d6fc, 0x01d714},
+ {0x01d716, 0x01d71b},
+ {0x01d736, 0x01d74e},
+ {0x01d750, 0x01d755},
+ {0x01d770, 0x01d788},
+ {0x01d78a, 0x01d78f},
+ {0x01d7aa, 0x01d7c2},
+ {0x01d7c4, 0x01d7c9},
+ {0x01d7cb, 0x01d7cb},
+ {0x01df00, 0x01df09},
+ {0x01df0b, 0x01df1e},
+ {0x01df25, 0x01df2a},
+ {0x01e030, 0x01e06d},
+ {0x01e922, 0x01e943}
+};
+
+/* table of Unicode codepoint ranges of Uppercase characters */
+static const pg_unicode_range unicode_uppercase[651] =
+{
+ {0x000041, 0x00005a},
+ {0x0000c0, 0x0000d6},
+ {0x0000d8, 0x0000de},
+ {0x000100, 0x000100},
+ {0x000102, 0x000102},
+ {0x000104, 0x000104},
+ {0x000106, 0x000106},
+ {0x000108, 0x000108},
+ {0x00010a, 0x00010a},
+ {0x00010c, 0x00010c},
+ {0x00010e, 0x00010e},
+ {0x000110, 0x000110},
+ {0x000112, 0x000112},
+ {0x000114, 0x000114},
+ {0x000116, 0x000116},
+ {0x000118, 0x000118},
+ {0x00011a, 0x00011a},
+ {0x00011c, 0x00011c},
+ {0x00011e, 0x00011e},
+ {0x000120, 0x000120},
+ {0x000122, 0x000122},
+ {0x000124, 0x000124},
+ {0x000126, 0x000126},
+ {0x000128, 0x000128},
+ {0x00012a, 0x00012a},
+ {0x00012c, 0x00012c},
+ {0x00012e, 0x00012e},
+ {0x000130, 0x000130},
+ {0x000132, 0x000132},
+ {0x000134, 0x000134},
+ {0x000136, 0x000136},
+ {0x000139, 0x000139},
+ {0x00013b, 0x00013b},
+ {0x00013d, 0x00013d},
+ {0x00013f, 0x00013f},
+ {0x000141, 0x000141},
+ {0x000143, 0x000143},
+ {0x000145, 0x000145},
+ {0x000147, 0x000147},
+ {0x00014a, 0x00014a},
+ {0x00014c, 0x00014c},
+ {0x00014e, 0x00014e},
+ {0x000150, 0x000150},
+ {0x000152, 0x000152},
+ {0x000154, 0x000154},
+ {0x000156, 0x000156},
+ {0x000158, 0x000158},
+ {0x00015a, 0x00015a},
+ {0x00015c, 0x00015c},
+ {0x00015e, 0x00015e},
+ {0x000160, 0x000160},
+ {0x000162, 0x000162},
+ {0x000164, 0x000164},
+ {0x000166, 0x000166},
+ {0x000168, 0x000168},
+ {0x00016a, 0x00016a},
+ {0x00016c, 0x00016c},
+ {0x00016e, 0x00016e},
+ {0x000170, 0x000170},
+ {0x000172, 0x000172},
+ {0x000174, 0x000174},
+ {0x000176, 0x000176},
+ {0x000178, 0x000179},
+ {0x00017b, 0x00017b},
+ {0x00017d, 0x00017d},
+ {0x000181, 0x000182},
+ {0x000184, 0x000184},
+ {0x000186, 0x000187},
+ {0x000189, 0x00018b},
+ {0x00018e, 0x000191},
+ {0x000193, 0x000194},
+ {0x000196, 0x000198},
+ {0x00019c, 0x00019d},
+ {0x00019f, 0x0001a0},
+ {0x0001a2, 0x0001a2},
+ {0x0001a4, 0x0001a4},
+ {0x0001a6, 0x0001a7},
+ {0x0001a9, 0x0001a9},
+ {0x0001ac, 0x0001ac},
+ {0x0001ae, 0x0001af},
+ {0x0001b1, 0x0001b3},
+ {0x0001b5, 0x0001b5},
+ {0x0001b7, 0x0001b8},
+ {0x0001bc, 0x0001bc},
+ {0x0001c4, 0x0001c4},
+ {0x0001c7, 0x0001c7},
+ {0x0001ca, 0x0001ca},
+ {0x0001cd, 0x0001cd},
+ {0x0001cf, 0x0001cf},
+ {0x0001d1, 0x0001d1},
+ {0x0001d3, 0x0001d3},
+ {0x0001d5, 0x0001d5},
+ {0x0001d7, 0x0001d7},
+ {0x0001d9, 0x0001d9},
+ {0x0001db, 0x0001db},
+ {0x0001de, 0x0001de},
+ {0x0001e0, 0x0001e0},
+ {0x0001e2, 0x0001e2},
+ {0x0001e4, 0x0001e4},
+ {0x0001e6, 0x0001e6},
+ {0x0001e8, 0x0001e8},
+ {0x0001ea, 0x0001ea},
+ {0x0001ec, 0x0001ec},
+ {0x0001ee, 0x0001ee},
+ {0x0001f1, 0x0001f1},
+ {0x0001f4, 0x0001f4},
+ {0x0001f6, 0x0001f8},
+ {0x0001fa, 0x0001fa},
+ {0x0001fc, 0x0001fc},
+ {0x0001fe, 0x0001fe},
+ {0x000200, 0x000200},
+ {0x000202, 0x000202},
+ {0x000204, 0x000204},
+ {0x000206, 0x000206},
+ {0x000208, 0x000208},
+ {0x00020a, 0x00020a},
+ {0x00020c, 0x00020c},
+ {0x00020e, 0x00020e},
+ {0x000210, 0x000210},
+ {0x000212, 0x000212},
+ {0x000214, 0x000214},
+ {0x000216, 0x000216},
+ {0x000218, 0x000218},
+ {0x00021a, 0x00021a},
+ {0x00021c, 0x00021c},
+ {0x00021e, 0x00021e},
+ {0x000220, 0x000220},
+ {0x000222, 0x000222},
+ {0x000224, 0x000224},
+ {0x000226, 0x000226},
+ {0x000228, 0x000228},
+ {0x00022a, 0x00022a},
+ {0x00022c, 0x00022c},
+ {0x00022e, 0x00022e},
+ {0x000230, 0x000230},
+ {0x000232, 0x000232},
+ {0x00023a, 0x00023b},
+ {0x00023d, 0x00023e},
+ {0x000241, 0x000241},
+ {0x000243, 0x000246},
+ {0x000248, 0x000248},
+ {0x00024a, 0x00024a},
+ {0x00024c, 0x00024c},
+ {0x00024e, 0x00024e},
+ {0x000370, 0x000370},
+ {0x000372, 0x000372},
+ {0x000376, 0x000376},
+ {0x00037f, 0x00037f},
+ {0x000386, 0x000386},
+ {0x000388, 0x00038a},
+ {0x00038c, 0x00038c},
+ {0x00038e, 0x00038f},
+ {0x000391, 0x0003a1},
+ {0x0003a3, 0x0003ab},
+ {0x0003cf, 0x0003cf},
+ {0x0003d2, 0x0003d4},
+ {0x0003d8, 0x0003d8},
+ {0x0003da, 0x0003da},
+ {0x0003dc, 0x0003dc},
+ {0x0003de, 0x0003de},
+ {0x0003e0, 0x0003e0},
+ {0x0003e2, 0x0003e2},
+ {0x0003e4, 0x0003e4},
+ {0x0003e6, 0x0003e6},
+ {0x0003e8, 0x0003e8},
+ {0x0003ea, 0x0003ea},
+ {0x0003ec, 0x0003ec},
+ {0x0003ee, 0x0003ee},
+ {0x0003f4, 0x0003f4},
+ {0x0003f7, 0x0003f7},
+ {0x0003f9, 0x0003fa},
+ {0x0003fd, 0x00042f},
+ {0x000460, 0x000460},
+ {0x000462, 0x000462},
+ {0x000464, 0x000464},
+ {0x000466, 0x000466},
+ {0x000468, 0x000468},
+ {0x00046a, 0x00046a},
+ {0x00046c, 0x00046c},
+ {0x00046e, 0x00046e},
+ {0x000470, 0x000470},
+ {0x000472, 0x000472},
+ {0x000474, 0x000474},
+ {0x000476, 0x000476},
+ {0x000478, 0x000478},
+ {0x00047a, 0x00047a},
+ {0x00047c, 0x00047c},
+ {0x00047e, 0x00047e},
+ {0x000480, 0x000480},
+ {0x00048a, 0x00048a},
+ {0x00048c, 0x00048c},
+ {0x00048e, 0x00048e},
+ {0x000490, 0x000490},
+ {0x000492, 0x000492},
+ {0x000494, 0x000494},
+ {0x000496, 0x000496},
+ {0x000498, 0x000498},
+ {0x00049a, 0x00049a},
+ {0x00049c, 0x00049c},
+ {0x00049e, 0x00049e},
+ {0x0004a0, 0x0004a0},
+ {0x0004a2, 0x0004a2},
+ {0x0004a4, 0x0004a4},
+ {0x0004a6, 0x0004a6},
+ {0x0004a8, 0x0004a8},
+ {0x0004aa, 0x0004aa},
+ {0x0004ac, 0x0004ac},
+ {0x0004ae, 0x0004ae},
+ {0x0004b0, 0x0004b0},
+ {0x0004b2, 0x0004b2},
+ {0x0004b4, 0x0004b4},
+ {0x0004b6, 0x0004b6},
+ {0x0004b8, 0x0004b8},
+ {0x0004ba, 0x0004ba},
+ {0x0004bc, 0x0004bc},
+ {0x0004be, 0x0004be},
+ {0x0004c0, 0x0004c1},
+ {0x0004c3, 0x0004c3},
+ {0x0004c5, 0x0004c5},
+ {0x0004c7, 0x0004c7},
+ {0x0004c9, 0x0004c9},
+ {0x0004cb, 0x0004cb},
+ {0x0004cd, 0x0004cd},
+ {0x0004d0, 0x0004d0},
+ {0x0004d2, 0x0004d2},
+ {0x0004d4, 0x0004d4},
+ {0x0004d6, 0x0004d6},
+ {0x0004d8, 0x0004d8},
+ {0x0004da, 0x0004da},
+ {0x0004dc, 0x0004dc},
+ {0x0004de, 0x0004de},
+ {0x0004e0, 0x0004e0},
+ {0x0004e2, 0x0004e2},
+ {0x0004e4, 0x0004e4},
+ {0x0004e6, 0x0004e6},
+ {0x0004e8, 0x0004e8},
+ {0x0004ea, 0x0004ea},
+ {0x0004ec, 0x0004ec},
+ {0x0004ee, 0x0004ee},
+ {0x0004f0, 0x0004f0},
+ {0x0004f2, 0x0004f2},
+ {0x0004f4, 0x0004f4},
+ {0x0004f6, 0x0004f6},
+ {0x0004f8, 0x0004f8},
+ {0x0004fa, 0x0004fa},
+ {0x0004fc, 0x0004fc},
+ {0x0004fe, 0x0004fe},
+ {0x000500, 0x000500},
+ {0x000502, 0x000502},
+ {0x000504, 0x000504},
+ {0x000506, 0x000506},
+ {0x000508, 0x000508},
+ {0x00050a, 0x00050a},
+ {0x00050c, 0x00050c},
+ {0x00050e, 0x00050e},
+ {0x000510, 0x000510},
+ {0x000512, 0x000512},
+ {0x000514, 0x000514},
+ {0x000516, 0x000516},
+ {0x000518, 0x000518},
+ {0x00051a, 0x00051a},
+ {0x00051c, 0x00051c},
+ {0x00051e, 0x00051e},
+ {0x000520, 0x000520},
+ {0x000522, 0x000522},
+ {0x000524, 0x000524},
+ {0x000526, 0x000526},
+ {0x000528, 0x000528},
+ {0x00052a, 0x00052a},
+ {0x00052c, 0x00052c},
+ {0x00052e, 0x00052e},
+ {0x000531, 0x000556},
+ {0x0010a0, 0x0010c5},
+ {0x0010c7, 0x0010c7},
+ {0x0010cd, 0x0010cd},
+ {0x0013a0, 0x0013f5},
+ {0x001c90, 0x001cba},
+ {0x001cbd, 0x001cbf},
+ {0x001e00, 0x001e00},
+ {0x001e02, 0x001e02},
+ {0x001e04, 0x001e04},
+ {0x001e06, 0x001e06},
+ {0x001e08, 0x001e08},
+ {0x001e0a, 0x001e0a},
+ {0x001e0c, 0x001e0c},
+ {0x001e0e, 0x001e0e},
+ {0x001e10, 0x001e10},
+ {0x001e12, 0x001e12},
+ {0x001e14, 0x001e14},
+ {0x001e16, 0x001e16},
+ {0x001e18, 0x001e18},
+ {0x001e1a, 0x001e1a},
+ {0x001e1c, 0x001e1c},
+ {0x001e1e, 0x001e1e},
+ {0x001e20, 0x001e20},
+ {0x001e22, 0x001e22},
+ {0x001e24, 0x001e24},
+ {0x001e26, 0x001e26},
+ {0x001e28, 0x001e28},
+ {0x001e2a, 0x001e2a},
+ {0x001e2c, 0x001e2c},
+ {0x001e2e, 0x001e2e},
+ {0x001e30, 0x001e30},
+ {0x001e32, 0x001e32},
+ {0x001e34, 0x001e34},
+ {0x001e36, 0x001e36},
+ {0x001e38, 0x001e38},
+ {0x001e3a, 0x001e3a},
+ {0x001e3c, 0x001e3c},
+ {0x001e3e, 0x001e3e},
+ {0x001e40, 0x001e40},
+ {0x001e42, 0x001e42},
+ {0x001e44, 0x001e44},
+ {0x001e46, 0x001e46},
+ {0x001e48, 0x001e48},
+ {0x001e4a, 0x001e4a},
+ {0x001e4c, 0x001e4c},
+ {0x001e4e, 0x001e4e},
+ {0x001e50, 0x001e50},
+ {0x001e52, 0x001e52},
+ {0x001e54, 0x001e54},
+ {0x001e56, 0x001e56},
+ {0x001e58, 0x001e58},
+ {0x001e5a, 0x001e5a},
+ {0x001e5c, 0x001e5c},
+ {0x001e5e, 0x001e5e},
+ {0x001e60, 0x001e60},
+ {0x001e62, 0x001e62},
+ {0x001e64, 0x001e64},
+ {0x001e66, 0x001e66},
+ {0x001e68, 0x001e68},
+ {0x001e6a, 0x001e6a},
+ {0x001e6c, 0x001e6c},
+ {0x001e6e, 0x001e6e},
+ {0x001e70, 0x001e70},
+ {0x001e72, 0x001e72},
+ {0x001e74, 0x001e74},
+ {0x001e76, 0x001e76},
+ {0x001e78, 0x001e78},
+ {0x001e7a, 0x001e7a},
+ {0x001e7c, 0x001e7c},
+ {0x001e7e, 0x001e7e},
+ {0x001e80, 0x001e80},
+ {0x001e82, 0x001e82},
+ {0x001e84, 0x001e84},
+ {0x001e86, 0x001e86},
+ {0x001e88, 0x001e88},
+ {0x001e8a, 0x001e8a},
+ {0x001e8c, 0x001e8c},
+ {0x001e8e, 0x001e8e},
+ {0x001e90, 0x001e90},
+ {0x001e92, 0x001e92},
+ {0x001e94, 0x001e94},
+ {0x001e9e, 0x001e9e},
+ {0x001ea0, 0x001ea0},
+ {0x001ea2, 0x001ea2},
+ {0x001ea4, 0x001ea4},
+ {0x001ea6, 0x001ea6},
+ {0x001ea8, 0x001ea8},
+ {0x001eaa, 0x001eaa},
+ {0x001eac, 0x001eac},
+ {0x001eae, 0x001eae},
+ {0x001eb0, 0x001eb0},
+ {0x001eb2, 0x001eb2},
+ {0x001eb4, 0x001eb4},
+ {0x001eb6, 0x001eb6},
+ {0x001eb8, 0x001eb8},
+ {0x001eba, 0x001eba},
+ {0x001ebc, 0x001ebc},
+ {0x001ebe, 0x001ebe},
+ {0x001ec0, 0x001ec0},
+ {0x001ec2, 0x001ec2},
+ {0x001ec4, 0x001ec4},
+ {0x001ec6, 0x001ec6},
+ {0x001ec8, 0x001ec8},
+ {0x001eca, 0x001eca},
+ {0x001ecc, 0x001ecc},
+ {0x001ece, 0x001ece},
+ {0x001ed0, 0x001ed0},
+ {0x001ed2, 0x001ed2},
+ {0x001ed4, 0x001ed4},
+ {0x001ed6, 0x001ed6},
+ {0x001ed8, 0x001ed8},
+ {0x001eda, 0x001eda},
+ {0x001edc, 0x001edc},
+ {0x001ede, 0x001ede},
+ {0x001ee0, 0x001ee0},
+ {0x001ee2, 0x001ee2},
+ {0x001ee4, 0x001ee4},
+ {0x001ee6, 0x001ee6},
+ {0x001ee8, 0x001ee8},
+ {0x001eea, 0x001eea},
+ {0x001eec, 0x001eec},
+ {0x001eee, 0x001eee},
+ {0x001ef0, 0x001ef0},
+ {0x001ef2, 0x001ef2},
+ {0x001ef4, 0x001ef4},
+ {0x001ef6, 0x001ef6},
+ {0x001ef8, 0x001ef8},
+ {0x001efa, 0x001efa},
+ {0x001efc, 0x001efc},
+ {0x001efe, 0x001efe},
+ {0x001f08, 0x001f0f},
+ {0x001f18, 0x001f1d},
+ {0x001f28, 0x001f2f},
+ {0x001f38, 0x001f3f},
+ {0x001f48, 0x001f4d},
+ {0x001f59, 0x001f59},
+ {0x001f5b, 0x001f5b},
+ {0x001f5d, 0x001f5d},
+ {0x001f5f, 0x001f5f},
+ {0x001f68, 0x001f6f},
+ {0x001fb8, 0x001fbb},
+ {0x001fc8, 0x001fcb},
+ {0x001fd8, 0x001fdb},
+ {0x001fe8, 0x001fec},
+ {0x001ff8, 0x001ffb},
+ {0x002102, 0x002102},
+ {0x002107, 0x002107},
+ {0x00210b, 0x00210d},
+ {0x002110, 0x002112},
+ {0x002115, 0x002115},
+ {0x002119, 0x00211d},
+ {0x002124, 0x002124},
+ {0x002126, 0x002126},
+ {0x002128, 0x002128},
+ {0x00212a, 0x00212d},
+ {0x002130, 0x002133},
+ {0x00213e, 0x00213f},
+ {0x002145, 0x002145},
+ {0x002160, 0x00216f},
+ {0x002183, 0x002183},
+ {0x0024b6, 0x0024cf},
+ {0x002c00, 0x002c2f},
+ {0x002c60, 0x002c60},
+ {0x002c62, 0x002c64},
+ {0x002c67, 0x002c67},
+ {0x002c69, 0x002c69},
+ {0x002c6b, 0x002c6b},
+ {0x002c6d, 0x002c70},
+ {0x002c72, 0x002c72},
+ {0x002c75, 0x002c75},
+ {0x002c7e, 0x002c80},
+ {0x002c82, 0x002c82},
+ {0x002c84, 0x002c84},
+ {0x002c86, 0x002c86},
+ {0x002c88, 0x002c88},
+ {0x002c8a, 0x002c8a},
+ {0x002c8c, 0x002c8c},
+ {0x002c8e, 0x002c8e},
+ {0x002c90, 0x002c90},
+ {0x002c92, 0x002c92},
+ {0x002c94, 0x002c94},
+ {0x002c96, 0x002c96},
+ {0x002c98, 0x002c98},
+ {0x002c9a, 0x002c9a},
+ {0x002c9c, 0x002c9c},
+ {0x002c9e, 0x002c9e},
+ {0x002ca0, 0x002ca0},
+ {0x002ca2, 0x002ca2},
+ {0x002ca4, 0x002ca4},
+ {0x002ca6, 0x002ca6},
+ {0x002ca8, 0x002ca8},
+ {0x002caa, 0x002caa},
+ {0x002cac, 0x002cac},
+ {0x002cae, 0x002cae},
+ {0x002cb0, 0x002cb0},
+ {0x002cb2, 0x002cb2},
+ {0x002cb4, 0x002cb4},
+ {0x002cb6, 0x002cb6},
+ {0x002cb8, 0x002cb8},
+ {0x002cba, 0x002cba},
+ {0x002cbc, 0x002cbc},
+ {0x002cbe, 0x002cbe},
+ {0x002cc0, 0x002cc0},
+ {0x002cc2, 0x002cc2},
+ {0x002cc4, 0x002cc4},
+ {0x002cc6, 0x002cc6},
+ {0x002cc8, 0x002cc8},
+ {0x002cca, 0x002cca},
+ {0x002ccc, 0x002ccc},
+ {0x002cce, 0x002cce},
+ {0x002cd0, 0x002cd0},
+ {0x002cd2, 0x002cd2},
+ {0x002cd4, 0x002cd4},
+ {0x002cd6, 0x002cd6},
+ {0x002cd8, 0x002cd8},
+ {0x002cda, 0x002cda},
+ {0x002cdc, 0x002cdc},
+ {0x002cde, 0x002cde},
+ {0x002ce0, 0x002ce0},
+ {0x002ce2, 0x002ce2},
+ {0x002ceb, 0x002ceb},
+ {0x002ced, 0x002ced},
+ {0x002cf2, 0x002cf2},
+ {0x00a640, 0x00a640},
+ {0x00a642, 0x00a642},
+ {0x00a644, 0x00a644},
+ {0x00a646, 0x00a646},
+ {0x00a648, 0x00a648},
+ {0x00a64a, 0x00a64a},
+ {0x00a64c, 0x00a64c},
+ {0x00a64e, 0x00a64e},
+ {0x00a650, 0x00a650},
+ {0x00a652, 0x00a652},
+ {0x00a654, 0x00a654},
+ {0x00a656, 0x00a656},
+ {0x00a658, 0x00a658},
+ {0x00a65a, 0x00a65a},
+ {0x00a65c, 0x00a65c},
+ {0x00a65e, 0x00a65e},
+ {0x00a660, 0x00a660},
+ {0x00a662, 0x00a662},
+ {0x00a664, 0x00a664},
+ {0x00a666, 0x00a666},
+ {0x00a668, 0x00a668},
+ {0x00a66a, 0x00a66a},
+ {0x00a66c, 0x00a66c},
+ {0x00a680, 0x00a680},
+ {0x00a682, 0x00a682},
+ {0x00a684, 0x00a684},
+ {0x00a686, 0x00a686},
+ {0x00a688, 0x00a688},
+ {0x00a68a, 0x00a68a},
+ {0x00a68c, 0x00a68c},
+ {0x00a68e, 0x00a68e},
+ {0x00a690, 0x00a690},
+ {0x00a692, 0x00a692},
+ {0x00a694, 0x00a694},
+ {0x00a696, 0x00a696},
+ {0x00a698, 0x00a698},
+ {0x00a69a, 0x00a69a},
+ {0x00a722, 0x00a722},
+ {0x00a724, 0x00a724},
+ {0x00a726, 0x00a726},
+ {0x00a728, 0x00a728},
+ {0x00a72a, 0x00a72a},
+ {0x00a72c, 0x00a72c},
+ {0x00a72e, 0x00a72e},
+ {0x00a732, 0x00a732},
+ {0x00a734, 0x00a734},
+ {0x00a736, 0x00a736},
+ {0x00a738, 0x00a738},
+ {0x00a73a, 0x00a73a},
+ {0x00a73c, 0x00a73c},
+ {0x00a73e, 0x00a73e},
+ {0x00a740, 0x00a740},
+ {0x00a742, 0x00a742},
+ {0x00a744, 0x00a744},
+ {0x00a746, 0x00a746},
+ {0x00a748, 0x00a748},
+ {0x00a74a, 0x00a74a},
+ {0x00a74c, 0x00a74c},
+ {0x00a74e, 0x00a74e},
+ {0x00a750, 0x00a750},
+ {0x00a752, 0x00a752},
+ {0x00a754, 0x00a754},
+ {0x00a756, 0x00a756},
+ {0x00a758, 0x00a758},
+ {0x00a75a, 0x00a75a},
+ {0x00a75c, 0x00a75c},
+ {0x00a75e, 0x00a75e},
+ {0x00a760, 0x00a760},
+ {0x00a762, 0x00a762},
+ {0x00a764, 0x00a764},
+ {0x00a766, 0x00a766},
+ {0x00a768, 0x00a768},
+ {0x00a76a, 0x00a76a},
+ {0x00a76c, 0x00a76c},
+ {0x00a76e, 0x00a76e},
+ {0x00a779, 0x00a779},
+ {0x00a77b, 0x00a77b},
+ {0x00a77d, 0x00a77e},
+ {0x00a780, 0x00a780},
+ {0x00a782, 0x00a782},
+ {0x00a784, 0x00a784},
+ {0x00a786, 0x00a786},
+ {0x00a78b, 0x00a78b},
+ {0x00a78d, 0x00a78d},
+ {0x00a790, 0x00a790},
+ {0x00a792, 0x00a792},
+ {0x00a796, 0x00a796},
+ {0x00a798, 0x00a798},
+ {0x00a79a, 0x00a79a},
+ {0x00a79c, 0x00a79c},
+ {0x00a79e, 0x00a79e},
+ {0x00a7a0, 0x00a7a0},
+ {0x00a7a2, 0x00a7a2},
+ {0x00a7a4, 0x00a7a4},
+ {0x00a7a6, 0x00a7a6},
+ {0x00a7a8, 0x00a7a8},
+ {0x00a7aa, 0x00a7ae},
+ {0x00a7b0, 0x00a7b4},
+ {0x00a7b6, 0x00a7b6},
+ {0x00a7b8, 0x00a7b8},
+ {0x00a7ba, 0x00a7ba},
+ {0x00a7bc, 0x00a7bc},
+ {0x00a7be, 0x00a7be},
+ {0x00a7c0, 0x00a7c0},
+ {0x00a7c2, 0x00a7c2},
+ {0x00a7c4, 0x00a7c7},
+ {0x00a7c9, 0x00a7c9},
+ {0x00a7d0, 0x00a7d0},
+ {0x00a7d6, 0x00a7d6},
+ {0x00a7d8, 0x00a7d8},
+ {0x00a7f5, 0x00a7f5},
+ {0x00ff21, 0x00ff3a},
+ {0x010400, 0x010427},
+ {0x0104b0, 0x0104d3},
+ {0x010570, 0x01057a},
+ {0x01057c, 0x01058a},
+ {0x01058c, 0x010592},
+ {0x010594, 0x010595},
+ {0x010c80, 0x010cb2},
+ {0x0118a0, 0x0118bf},
+ {0x016e40, 0x016e5f},
+ {0x01d400, 0x01d419},
+ {0x01d434, 0x01d44d},
+ {0x01d468, 0x01d481},
+ {0x01d49c, 0x01d49c},
+ {0x01d49e, 0x01d49f},
+ {0x01d4a2, 0x01d4a2},
+ {0x01d4a5, 0x01d4a6},
+ {0x01d4a9, 0x01d4ac},
+ {0x01d4ae, 0x01d4b5},
+ {0x01d4d0, 0x01d4e9},
+ {0x01d504, 0x01d505},
+ {0x01d507, 0x01d50a},
+ {0x01d50d, 0x01d514},
+ {0x01d516, 0x01d51c},
+ {0x01d538, 0x01d539},
+ {0x01d53b, 0x01d53e},
+ {0x01d540, 0x01d544},
+ {0x01d546, 0x01d546},
+ {0x01d54a, 0x01d550},
+ {0x01d56c, 0x01d585},
+ {0x01d5a0, 0x01d5b9},
+ {0x01d5d4, 0x01d5ed},
+ {0x01d608, 0x01d621},
+ {0x01d63c, 0x01d655},
+ {0x01d670, 0x01d689},
+ {0x01d6a8, 0x01d6c0},
+ {0x01d6e2, 0x01d6fa},
+ {0x01d71c, 0x01d734},
+ {0x01d756, 0x01d76e},
+ {0x01d790, 0x01d7a8},
+ {0x01d7ca, 0x01d7ca},
+ {0x01e900, 0x01e921},
+ {0x01f130, 0x01f149},
+ {0x01f150, 0x01f169},
+ {0x01f170, 0x01f189}
+};
+
+/* table of Unicode codepoint ranges of White_Space characters */
+static const pg_unicode_range unicode_white_space[11] =
+{
+ {0x000009, 0x00000d},
+ {0x000020, 0x000020},
+ {0x000085, 0x000085},
+ {0x0000a0, 0x0000a0},
+ {0x001680, 0x001680},
+ {0x002000, 0x00200a},
+ {0x002028, 0x002028},
+ {0x002029, 0x002029},
+ {0x00202f, 0x00202f},
+ {0x00205f, 0x00205f},
+ {0x003000, 0x003000}
+};
+
+/* table of Unicode codepoint ranges of Hex_Digit characters */
+static const pg_unicode_range unicode_hex_digit[6] =
+{
+ {0x000030, 0x000039},
+ {0x000041, 0x000046},
+ {0x000061, 0x000066},
+ {0x00ff10, 0x00ff19},
+ {0x00ff21, 0x00ff26},
+ {0x00ff41, 0x00ff46}
+};
+
+/* table of Unicode codepoint ranges of Join_Control characters */
+static const pg_unicode_range unicode_join_control[1] =
+{
+ {0x00200c, 0x00200d}
+};
--
2.34.1
[text/x-patch] v1-0002-Shrink-unicode-category-table.patch (101.7K, ../../[email protected]/3-v1-0002-Shrink-unicode-category-table.patch)
download | inline diff:
From 35cd57cd65205573be3a3eff91affe307da405d0 Mon Sep 17 00:00:00 2001
From: Jeff Davis <[email protected]>
Date: Wed, 22 Nov 2023 11:30:31 -0800
Subject: [PATCH v1 2/3] Shrink unicode category table.
Missing entries can implicitly be considered "unassigned".
---
.../generate-unicode_category_table.pl | 21 +-
src/common/unicode_category.c | 6 +-
src/include/common/unicode_category_table.h | 711 +-----------------
3 files changed, 15 insertions(+), 723 deletions(-)
diff --git a/src/common/unicode/generate-unicode_category_table.pl b/src/common/unicode/generate-unicode_category_table.pl
index 8f03425e0b..992b877ede 100644
--- a/src/common/unicode/generate-unicode_category_table.pl
+++ b/src/common/unicode/generate-unicode_category_table.pl
@@ -72,7 +72,10 @@ while (my $line = <$FH>)
# the current range, emit the current range and initialize a new
# range representing the gap.
if ($range_end + 1 != $code && $range_category ne $gap_category) {
- push(@category_ranges, {start => $range_start, end => $range_end, category => $range_category});
+ if ($range_category ne $CATEGORY_UNASSIGNED) {
+ push(@category_ranges, {start => $range_start, end => $range_end,
+ category => $range_category});
+ }
$range_start = $range_end + 1;
$range_end = $code - 1;
$range_category = $gap_category;
@@ -80,7 +83,10 @@ while (my $line = <$FH>)
# different category; new range
if ($range_category ne $category) {
- push(@category_ranges, {start => $range_start, end => $range_end, category => $range_category});
+ if ($range_category ne $CATEGORY_UNASSIGNED) {
+ push(@category_ranges, {start => $range_start, end => $range_end,
+ category => $range_category});
+ }
$range_start = $code;
$range_end = $code;
$range_category = $category;
@@ -109,14 +115,9 @@ die "<..., First> entry with no corresponding <..., Last> entry"
if $gap_category ne $CATEGORY_UNASSIGNED;
# emit final range
-push(@category_ranges, {start => $range_start, end => $range_end, category => $range_category});
-
-# emit range for any unassigned code points after last entry
-if ($range_end < 0x10FFFF) {
- $range_start = $range_end + 1;
- $range_end = 0x10FFFF;
- $range_category = $CATEGORY_UNASSIGNED;
- push(@category_ranges, {start => $range_start, end => $range_end, category => $range_category});
+if ($range_category ne $CATEGORY_UNASSIGNED) {
+ push(@category_ranges, {start => $range_start, end => $range_end,
+ category => $range_category});
}
my $num_ranges = scalar @category_ranges;
diff --git a/src/common/unicode_category.c b/src/common/unicode_category.c
index cec9c0d998..189cd6eca3 100644
--- a/src/common/unicode_category.c
+++ b/src/common/unicode_category.c
@@ -28,8 +28,7 @@ unicode_category(pg_wchar ucs)
int mid;
int max = lengthof(unicode_categories) - 1;
- Assert(ucs >= unicode_categories[0].first &&
- ucs <= unicode_categories[max].last);
+ Assert(ucs <= 0x10ffff);
while (max >= min)
{
@@ -42,8 +41,7 @@ unicode_category(pg_wchar ucs)
return unicode_categories[mid].category;
}
- Assert(false);
- return (pg_unicode_category) - 1;
+ return PG_U_UNASSIGNED;
}
/*
diff --git a/src/include/common/unicode_category_table.h b/src/include/common/unicode_category_table.h
index 06ad50d215..14f1ea0677 100644
--- a/src/include/common/unicode_category_table.h
+++ b/src/include/common/unicode_category_table.h
@@ -26,7 +26,7 @@ typedef struct
} pg_category_range;
/* table of Unicode codepoint ranges and their categories */
-static const pg_category_range unicode_categories[4009] =
+static const pg_category_range unicode_categories[3302] =
{
{0x000000, 0x00001f, PG_U_CONTROL},
{0x000020, 0x000020, PG_U_SPACE_SEPARATOR},
@@ -397,23 +397,18 @@ static const pg_category_range unicode_categories[4009] =
{0x000375, 0x000375, PG_U_MODIFIER_SYMBOL},
{0x000376, 0x000376, PG_U_UPPERCASE_LETTER},
{0x000377, 0x000377, PG_U_LOWERCASE_LETTER},
- {0x000378, 0x000379, PG_U_UNASSIGNED},
{0x00037a, 0x00037a, PG_U_MODIFIER_LETTER},
{0x00037b, 0x00037d, PG_U_LOWERCASE_LETTER},
{0x00037e, 0x00037e, PG_U_OTHER_PUNCTUATION},
{0x00037f, 0x00037f, PG_U_UPPERCASE_LETTER},
- {0x000380, 0x000383, PG_U_UNASSIGNED},
{0x000384, 0x000385, PG_U_MODIFIER_SYMBOL},
{0x000386, 0x000386, PG_U_UPPERCASE_LETTER},
{0x000387, 0x000387, PG_U_OTHER_PUNCTUATION},
{0x000388, 0x00038a, PG_U_UPPERCASE_LETTER},
- {0x00038b, 0x00038b, PG_U_UNASSIGNED},
{0x00038c, 0x00038c, PG_U_UPPERCASE_LETTER},
- {0x00038d, 0x00038d, PG_U_UNASSIGNED},
{0x00038e, 0x00038f, PG_U_UPPERCASE_LETTER},
{0x000390, 0x000390, PG_U_LOWERCASE_LETTER},
{0x000391, 0x0003a1, PG_U_UPPERCASE_LETTER},
- {0x0003a2, 0x0003a2, PG_U_UNASSIGNED},
{0x0003a3, 0x0003ab, PG_U_UPPERCASE_LETTER},
{0x0003ac, 0x0003ce, PG_U_LOWERCASE_LETTER},
{0x0003cf, 0x0003cf, PG_U_UPPERCASE_LETTER},
@@ -654,18 +649,14 @@ static const pg_category_range unicode_categories[4009] =
{0x00052d, 0x00052d, PG_U_LOWERCASE_LETTER},
{0x00052e, 0x00052e, PG_U_UPPERCASE_LETTER},
{0x00052f, 0x00052f, PG_U_LOWERCASE_LETTER},
- {0x000530, 0x000530, PG_U_UNASSIGNED},
{0x000531, 0x000556, PG_U_UPPERCASE_LETTER},
- {0x000557, 0x000558, PG_U_UNASSIGNED},
{0x000559, 0x000559, PG_U_MODIFIER_LETTER},
{0x00055a, 0x00055f, PG_U_OTHER_PUNCTUATION},
{0x000560, 0x000588, PG_U_LOWERCASE_LETTER},
{0x000589, 0x000589, PG_U_OTHER_PUNCTUATION},
{0x00058a, 0x00058a, PG_U_DASH_PUNCTUATION},
- {0x00058b, 0x00058c, PG_U_UNASSIGNED},
{0x00058d, 0x00058e, PG_U_OTHER_SYMBOL},
{0x00058f, 0x00058f, PG_U_CURRENCY_SYMBOL},
- {0x000590, 0x000590, PG_U_UNASSIGNED},
{0x000591, 0x0005bd, PG_U_NONSPACING_MARK},
{0x0005be, 0x0005be, PG_U_DASH_PUNCTUATION},
{0x0005bf, 0x0005bf, PG_U_NONSPACING_MARK},
@@ -675,12 +666,9 @@ static const pg_category_range unicode_categories[4009] =
{0x0005c4, 0x0005c5, PG_U_NONSPACING_MARK},
{0x0005c6, 0x0005c6, PG_U_OTHER_PUNCTUATION},
{0x0005c7, 0x0005c7, PG_U_NONSPACING_MARK},
- {0x0005c8, 0x0005cf, PG_U_UNASSIGNED},
{0x0005d0, 0x0005ea, PG_U_OTHER_LETTER},
- {0x0005eb, 0x0005ee, PG_U_UNASSIGNED},
{0x0005ef, 0x0005f2, PG_U_OTHER_LETTER},
{0x0005f3, 0x0005f4, PG_U_OTHER_PUNCTUATION},
- {0x0005f5, 0x0005ff, PG_U_UNASSIGNED},
{0x000600, 0x000605, PG_U_FORMAT},
{0x000606, 0x000608, PG_U_MATH_SYMBOL},
{0x000609, 0x00060a, PG_U_OTHER_PUNCTUATION},
@@ -716,17 +704,14 @@ static const pg_category_range unicode_categories[4009] =
{0x0006fd, 0x0006fe, PG_U_OTHER_SYMBOL},
{0x0006ff, 0x0006ff, PG_U_OTHER_LETTER},
{0x000700, 0x00070d, PG_U_OTHER_PUNCTUATION},
- {0x00070e, 0x00070e, PG_U_UNASSIGNED},
{0x00070f, 0x00070f, PG_U_FORMAT},
{0x000710, 0x000710, PG_U_OTHER_LETTER},
{0x000711, 0x000711, PG_U_NONSPACING_MARK},
{0x000712, 0x00072f, PG_U_OTHER_LETTER},
{0x000730, 0x00074a, PG_U_NONSPACING_MARK},
- {0x00074b, 0x00074c, PG_U_UNASSIGNED},
{0x00074d, 0x0007a5, PG_U_OTHER_LETTER},
{0x0007a6, 0x0007b0, PG_U_NONSPACING_MARK},
{0x0007b1, 0x0007b1, PG_U_OTHER_LETTER},
- {0x0007b2, 0x0007bf, PG_U_UNASSIGNED},
{0x0007c0, 0x0007c9, PG_U_DECIMAL_NUMBER},
{0x0007ca, 0x0007ea, PG_U_OTHER_LETTER},
{0x0007eb, 0x0007f3, PG_U_NONSPACING_MARK},
@@ -734,7 +719,6 @@ static const pg_category_range unicode_categories[4009] =
{0x0007f6, 0x0007f6, PG_U_OTHER_SYMBOL},
{0x0007f7, 0x0007f9, PG_U_OTHER_PUNCTUATION},
{0x0007fa, 0x0007fa, PG_U_MODIFIER_LETTER},
- {0x0007fb, 0x0007fc, PG_U_UNASSIGNED},
{0x0007fd, 0x0007fd, PG_U_NONSPACING_MARK},
{0x0007fe, 0x0007ff, PG_U_CURRENCY_SYMBOL},
{0x000800, 0x000815, PG_U_OTHER_LETTER},
@@ -745,22 +729,15 @@ static const pg_category_range unicode_categories[4009] =
{0x000825, 0x000827, PG_U_NONSPACING_MARK},
{0x000828, 0x000828, PG_U_MODIFIER_LETTER},
{0x000829, 0x00082d, PG_U_NONSPACING_MARK},
- {0x00082e, 0x00082f, PG_U_UNASSIGNED},
{0x000830, 0x00083e, PG_U_OTHER_PUNCTUATION},
- {0x00083f, 0x00083f, PG_U_UNASSIGNED},
{0x000840, 0x000858, PG_U_OTHER_LETTER},
{0x000859, 0x00085b, PG_U_NONSPACING_MARK},
- {0x00085c, 0x00085d, PG_U_UNASSIGNED},
{0x00085e, 0x00085e, PG_U_OTHER_PUNCTUATION},
- {0x00085f, 0x00085f, PG_U_UNASSIGNED},
{0x000860, 0x00086a, PG_U_OTHER_LETTER},
- {0x00086b, 0x00086f, PG_U_UNASSIGNED},
{0x000870, 0x000887, PG_U_OTHER_LETTER},
{0x000888, 0x000888, PG_U_MODIFIER_SYMBOL},
{0x000889, 0x00088e, PG_U_OTHER_LETTER},
- {0x00088f, 0x00088f, PG_U_UNASSIGNED},
{0x000890, 0x000891, PG_U_FORMAT},
- {0x000892, 0x000897, PG_U_UNASSIGNED},
{0x000898, 0x00089f, PG_U_NONSPACING_MARK},
{0x0008a0, 0x0008c8, PG_U_OTHER_LETTER},
{0x0008c9, 0x0008c9, PG_U_MODIFIER_LETTER},
@@ -789,37 +766,24 @@ static const pg_category_range unicode_categories[4009] =
{0x000972, 0x000980, PG_U_OTHER_LETTER},
{0x000981, 0x000981, PG_U_NONSPACING_MARK},
{0x000982, 0x000983, PG_U_SPACING_MARK},
- {0x000984, 0x000984, PG_U_UNASSIGNED},
{0x000985, 0x00098c, PG_U_OTHER_LETTER},
- {0x00098d, 0x00098e, PG_U_UNASSIGNED},
{0x00098f, 0x000990, PG_U_OTHER_LETTER},
- {0x000991, 0x000992, PG_U_UNASSIGNED},
{0x000993, 0x0009a8, PG_U_OTHER_LETTER},
- {0x0009a9, 0x0009a9, PG_U_UNASSIGNED},
{0x0009aa, 0x0009b0, PG_U_OTHER_LETTER},
- {0x0009b1, 0x0009b1, PG_U_UNASSIGNED},
{0x0009b2, 0x0009b2, PG_U_OTHER_LETTER},
- {0x0009b3, 0x0009b5, PG_U_UNASSIGNED},
{0x0009b6, 0x0009b9, PG_U_OTHER_LETTER},
- {0x0009ba, 0x0009bb, PG_U_UNASSIGNED},
{0x0009bc, 0x0009bc, PG_U_NONSPACING_MARK},
{0x0009bd, 0x0009bd, PG_U_OTHER_LETTER},
{0x0009be, 0x0009c0, PG_U_SPACING_MARK},
{0x0009c1, 0x0009c4, PG_U_NONSPACING_MARK},
- {0x0009c5, 0x0009c6, PG_U_UNASSIGNED},
{0x0009c7, 0x0009c8, PG_U_SPACING_MARK},
- {0x0009c9, 0x0009ca, PG_U_UNASSIGNED},
{0x0009cb, 0x0009cc, PG_U_SPACING_MARK},
{0x0009cd, 0x0009cd, PG_U_NONSPACING_MARK},
{0x0009ce, 0x0009ce, PG_U_OTHER_LETTER},
- {0x0009cf, 0x0009d6, PG_U_UNASSIGNED},
{0x0009d7, 0x0009d7, PG_U_SPACING_MARK},
- {0x0009d8, 0x0009db, PG_U_UNASSIGNED},
{0x0009dc, 0x0009dd, PG_U_OTHER_LETTER},
- {0x0009de, 0x0009de, PG_U_UNASSIGNED},
{0x0009df, 0x0009e1, PG_U_OTHER_LETTER},
{0x0009e2, 0x0009e3, PG_U_NONSPACING_MARK},
- {0x0009e4, 0x0009e5, PG_U_UNASSIGNED},
{0x0009e6, 0x0009ef, PG_U_DECIMAL_NUMBER},
{0x0009f0, 0x0009f1, PG_U_OTHER_LETTER},
{0x0009f2, 0x0009f3, PG_U_CURRENCY_SYMBOL},
@@ -829,194 +793,121 @@ static const pg_category_range unicode_categories[4009] =
{0x0009fc, 0x0009fc, PG_U_OTHER_LETTER},
{0x0009fd, 0x0009fd, PG_U_OTHER_PUNCTUATION},
{0x0009fe, 0x0009fe, PG_U_NONSPACING_MARK},
- {0x0009ff, 0x000a00, PG_U_UNASSIGNED},
{0x000a01, 0x000a02, PG_U_NONSPACING_MARK},
{0x000a03, 0x000a03, PG_U_SPACING_MARK},
- {0x000a04, 0x000a04, PG_U_UNASSIGNED},
{0x000a05, 0x000a0a, PG_U_OTHER_LETTER},
- {0x000a0b, 0x000a0e, PG_U_UNASSIGNED},
{0x000a0f, 0x000a10, PG_U_OTHER_LETTER},
- {0x000a11, 0x000a12, PG_U_UNASSIGNED},
{0x000a13, 0x000a28, PG_U_OTHER_LETTER},
- {0x000a29, 0x000a29, PG_U_UNASSIGNED},
{0x000a2a, 0x000a30, PG_U_OTHER_LETTER},
- {0x000a31, 0x000a31, PG_U_UNASSIGNED},
{0x000a32, 0x000a33, PG_U_OTHER_LETTER},
- {0x000a34, 0x000a34, PG_U_UNASSIGNED},
{0x000a35, 0x000a36, PG_U_OTHER_LETTER},
- {0x000a37, 0x000a37, PG_U_UNASSIGNED},
{0x000a38, 0x000a39, PG_U_OTHER_LETTER},
- {0x000a3a, 0x000a3b, PG_U_UNASSIGNED},
{0x000a3c, 0x000a3c, PG_U_NONSPACING_MARK},
- {0x000a3d, 0x000a3d, PG_U_UNASSIGNED},
{0x000a3e, 0x000a40, PG_U_SPACING_MARK},
{0x000a41, 0x000a42, PG_U_NONSPACING_MARK},
- {0x000a43, 0x000a46, PG_U_UNASSIGNED},
{0x000a47, 0x000a48, PG_U_NONSPACING_MARK},
- {0x000a49, 0x000a4a, PG_U_UNASSIGNED},
{0x000a4b, 0x000a4d, PG_U_NONSPACING_MARK},
- {0x000a4e, 0x000a50, PG_U_UNASSIGNED},
{0x000a51, 0x000a51, PG_U_NONSPACING_MARK},
- {0x000a52, 0x000a58, PG_U_UNASSIGNED},
{0x000a59, 0x000a5c, PG_U_OTHER_LETTER},
- {0x000a5d, 0x000a5d, PG_U_UNASSIGNED},
{0x000a5e, 0x000a5e, PG_U_OTHER_LETTER},
- {0x000a5f, 0x000a65, PG_U_UNASSIGNED},
{0x000a66, 0x000a6f, PG_U_DECIMAL_NUMBER},
{0x000a70, 0x000a71, PG_U_NONSPACING_MARK},
{0x000a72, 0x000a74, PG_U_OTHER_LETTER},
{0x000a75, 0x000a75, PG_U_NONSPACING_MARK},
{0x000a76, 0x000a76, PG_U_OTHER_PUNCTUATION},
- {0x000a77, 0x000a80, PG_U_UNASSIGNED},
{0x000a81, 0x000a82, PG_U_NONSPACING_MARK},
{0x000a83, 0x000a83, PG_U_SPACING_MARK},
- {0x000a84, 0x000a84, PG_U_UNASSIGNED},
{0x000a85, 0x000a8d, PG_U_OTHER_LETTER},
- {0x000a8e, 0x000a8e, PG_U_UNASSIGNED},
{0x000a8f, 0x000a91, PG_U_OTHER_LETTER},
- {0x000a92, 0x000a92, PG_U_UNASSIGNED},
{0x000a93, 0x000aa8, PG_U_OTHER_LETTER},
- {0x000aa9, 0x000aa9, PG_U_UNASSIGNED},
{0x000aaa, 0x000ab0, PG_U_OTHER_LETTER},
- {0x000ab1, 0x000ab1, PG_U_UNASSIGNED},
{0x000ab2, 0x000ab3, PG_U_OTHER_LETTER},
- {0x000ab4, 0x000ab4, PG_U_UNASSIGNED},
{0x000ab5, 0x000ab9, PG_U_OTHER_LETTER},
- {0x000aba, 0x000abb, PG_U_UNASSIGNED},
{0x000abc, 0x000abc, PG_U_NONSPACING_MARK},
{0x000abd, 0x000abd, PG_U_OTHER_LETTER},
{0x000abe, 0x000ac0, PG_U_SPACING_MARK},
{0x000ac1, 0x000ac5, PG_U_NONSPACING_MARK},
- {0x000ac6, 0x000ac6, PG_U_UNASSIGNED},
{0x000ac7, 0x000ac8, PG_U_NONSPACING_MARK},
{0x000ac9, 0x000ac9, PG_U_SPACING_MARK},
- {0x000aca, 0x000aca, PG_U_UNASSIGNED},
{0x000acb, 0x000acc, PG_U_SPACING_MARK},
{0x000acd, 0x000acd, PG_U_NONSPACING_MARK},
- {0x000ace, 0x000acf, PG_U_UNASSIGNED},
{0x000ad0, 0x000ad0, PG_U_OTHER_LETTER},
- {0x000ad1, 0x000adf, PG_U_UNASSIGNED},
{0x000ae0, 0x000ae1, PG_U_OTHER_LETTER},
{0x000ae2, 0x000ae3, PG_U_NONSPACING_MARK},
- {0x000ae4, 0x000ae5, PG_U_UNASSIGNED},
{0x000ae6, 0x000aef, PG_U_DECIMAL_NUMBER},
{0x000af0, 0x000af0, PG_U_OTHER_PUNCTUATION},
{0x000af1, 0x000af1, PG_U_CURRENCY_SYMBOL},
- {0x000af2, 0x000af8, PG_U_UNASSIGNED},
{0x000af9, 0x000af9, PG_U_OTHER_LETTER},
{0x000afa, 0x000aff, PG_U_NONSPACING_MARK},
- {0x000b00, 0x000b00, PG_U_UNASSIGNED},
{0x000b01, 0x000b01, PG_U_NONSPACING_MARK},
{0x000b02, 0x000b03, PG_U_SPACING_MARK},
- {0x000b04, 0x000b04, PG_U_UNASSIGNED},
{0x000b05, 0x000b0c, PG_U_OTHER_LETTER},
- {0x000b0d, 0x000b0e, PG_U_UNASSIGNED},
{0x000b0f, 0x000b10, PG_U_OTHER_LETTER},
- {0x000b11, 0x000b12, PG_U_UNASSIGNED},
{0x000b13, 0x000b28, PG_U_OTHER_LETTER},
- {0x000b29, 0x000b29, PG_U_UNASSIGNED},
{0x000b2a, 0x000b30, PG_U_OTHER_LETTER},
- {0x000b31, 0x000b31, PG_U_UNASSIGNED},
{0x000b32, 0x000b33, PG_U_OTHER_LETTER},
- {0x000b34, 0x000b34, PG_U_UNASSIGNED},
{0x000b35, 0x000b39, PG_U_OTHER_LETTER},
- {0x000b3a, 0x000b3b, PG_U_UNASSIGNED},
{0x000b3c, 0x000b3c, PG_U_NONSPACING_MARK},
{0x000b3d, 0x000b3d, PG_U_OTHER_LETTER},
{0x000b3e, 0x000b3e, PG_U_SPACING_MARK},
{0x000b3f, 0x000b3f, PG_U_NONSPACING_MARK},
{0x000b40, 0x000b40, PG_U_SPACING_MARK},
{0x000b41, 0x000b44, PG_U_NONSPACING_MARK},
- {0x000b45, 0x000b46, PG_U_UNASSIGNED},
{0x000b47, 0x000b48, PG_U_SPACING_MARK},
- {0x000b49, 0x000b4a, PG_U_UNASSIGNED},
{0x000b4b, 0x000b4c, PG_U_SPACING_MARK},
{0x000b4d, 0x000b4d, PG_U_NONSPACING_MARK},
- {0x000b4e, 0x000b54, PG_U_UNASSIGNED},
{0x000b55, 0x000b56, PG_U_NONSPACING_MARK},
{0x000b57, 0x000b57, PG_U_SPACING_MARK},
- {0x000b58, 0x000b5b, PG_U_UNASSIGNED},
{0x000b5c, 0x000b5d, PG_U_OTHER_LETTER},
- {0x000b5e, 0x000b5e, PG_U_UNASSIGNED},
{0x000b5f, 0x000b61, PG_U_OTHER_LETTER},
{0x000b62, 0x000b63, PG_U_NONSPACING_MARK},
- {0x000b64, 0x000b65, PG_U_UNASSIGNED},
{0x000b66, 0x000b6f, PG_U_DECIMAL_NUMBER},
{0x000b70, 0x000b70, PG_U_OTHER_SYMBOL},
{0x000b71, 0x000b71, PG_U_OTHER_LETTER},
{0x000b72, 0x000b77, PG_U_OTHER_NUMBER},
- {0x000b78, 0x000b81, PG_U_UNASSIGNED},
{0x000b82, 0x000b82, PG_U_NONSPACING_MARK},
{0x000b83, 0x000b83, PG_U_OTHER_LETTER},
- {0x000b84, 0x000b84, PG_U_UNASSIGNED},
{0x000b85, 0x000b8a, PG_U_OTHER_LETTER},
- {0x000b8b, 0x000b8d, PG_U_UNASSIGNED},
{0x000b8e, 0x000b90, PG_U_OTHER_LETTER},
- {0x000b91, 0x000b91, PG_U_UNASSIGNED},
{0x000b92, 0x000b95, PG_U_OTHER_LETTER},
- {0x000b96, 0x000b98, PG_U_UNASSIGNED},
{0x000b99, 0x000b9a, PG_U_OTHER_LETTER},
- {0x000b9b, 0x000b9b, PG_U_UNASSIGNED},
{0x000b9c, 0x000b9c, PG_U_OTHER_LETTER},
- {0x000b9d, 0x000b9d, PG_U_UNASSIGNED},
{0x000b9e, 0x000b9f, PG_U_OTHER_LETTER},
- {0x000ba0, 0x000ba2, PG_U_UNASSIGNED},
{0x000ba3, 0x000ba4, PG_U_OTHER_LETTER},
- {0x000ba5, 0x000ba7, PG_U_UNASSIGNED},
{0x000ba8, 0x000baa, PG_U_OTHER_LETTER},
- {0x000bab, 0x000bad, PG_U_UNASSIGNED},
{0x000bae, 0x000bb9, PG_U_OTHER_LETTER},
- {0x000bba, 0x000bbd, PG_U_UNASSIGNED},
{0x000bbe, 0x000bbf, PG_U_SPACING_MARK},
{0x000bc0, 0x000bc0, PG_U_NONSPACING_MARK},
{0x000bc1, 0x000bc2, PG_U_SPACING_MARK},
- {0x000bc3, 0x000bc5, PG_U_UNASSIGNED},
{0x000bc6, 0x000bc8, PG_U_SPACING_MARK},
- {0x000bc9, 0x000bc9, PG_U_UNASSIGNED},
{0x000bca, 0x000bcc, PG_U_SPACING_MARK},
{0x000bcd, 0x000bcd, PG_U_NONSPACING_MARK},
- {0x000bce, 0x000bcf, PG_U_UNASSIGNED},
{0x000bd0, 0x000bd0, PG_U_OTHER_LETTER},
- {0x000bd1, 0x000bd6, PG_U_UNASSIGNED},
{0x000bd7, 0x000bd7, PG_U_SPACING_MARK},
- {0x000bd8, 0x000be5, PG_U_UNASSIGNED},
{0x000be6, 0x000bef, PG_U_DECIMAL_NUMBER},
{0x000bf0, 0x000bf2, PG_U_OTHER_NUMBER},
{0x000bf3, 0x000bf8, PG_U_OTHER_SYMBOL},
{0x000bf9, 0x000bf9, PG_U_CURRENCY_SYMBOL},
{0x000bfa, 0x000bfa, PG_U_OTHER_SYMBOL},
- {0x000bfb, 0x000bff, PG_U_UNASSIGNED},
{0x000c00, 0x000c00, PG_U_NONSPACING_MARK},
{0x000c01, 0x000c03, PG_U_SPACING_MARK},
{0x000c04, 0x000c04, PG_U_NONSPACING_MARK},
{0x000c05, 0x000c0c, PG_U_OTHER_LETTER},
- {0x000c0d, 0x000c0d, PG_U_UNASSIGNED},
{0x000c0e, 0x000c10, PG_U_OTHER_LETTER},
- {0x000c11, 0x000c11, PG_U_UNASSIGNED},
{0x000c12, 0x000c28, PG_U_OTHER_LETTER},
- {0x000c29, 0x000c29, PG_U_UNASSIGNED},
{0x000c2a, 0x000c39, PG_U_OTHER_LETTER},
- {0x000c3a, 0x000c3b, PG_U_UNASSIGNED},
{0x000c3c, 0x000c3c, PG_U_NONSPACING_MARK},
{0x000c3d, 0x000c3d, PG_U_OTHER_LETTER},
{0x000c3e, 0x000c40, PG_U_NONSPACING_MARK},
{0x000c41, 0x000c44, PG_U_SPACING_MARK},
- {0x000c45, 0x000c45, PG_U_UNASSIGNED},
{0x000c46, 0x000c48, PG_U_NONSPACING_MARK},
- {0x000c49, 0x000c49, PG_U_UNASSIGNED},
{0x000c4a, 0x000c4d, PG_U_NONSPACING_MARK},
- {0x000c4e, 0x000c54, PG_U_UNASSIGNED},
{0x000c55, 0x000c56, PG_U_NONSPACING_MARK},
- {0x000c57, 0x000c57, PG_U_UNASSIGNED},
{0x000c58, 0x000c5a, PG_U_OTHER_LETTER},
- {0x000c5b, 0x000c5c, PG_U_UNASSIGNED},
{0x000c5d, 0x000c5d, PG_U_OTHER_LETTER},
- {0x000c5e, 0x000c5f, PG_U_UNASSIGNED},
{0x000c60, 0x000c61, PG_U_OTHER_LETTER},
{0x000c62, 0x000c63, PG_U_NONSPACING_MARK},
- {0x000c64, 0x000c65, PG_U_UNASSIGNED},
{0x000c66, 0x000c6f, PG_U_DECIMAL_NUMBER},
- {0x000c70, 0x000c76, PG_U_UNASSIGNED},
{0x000c77, 0x000c77, PG_U_OTHER_PUNCTUATION},
{0x000c78, 0x000c7e, PG_U_OTHER_NUMBER},
{0x000c7f, 0x000c7f, PG_U_OTHER_SYMBOL},
@@ -1025,101 +916,68 @@ static const pg_category_range unicode_categories[4009] =
{0x000c82, 0x000c83, PG_U_SPACING_MARK},
{0x000c84, 0x000c84, PG_U_OTHER_PUNCTUATION},
{0x000c85, 0x000c8c, PG_U_OTHER_LETTER},
- {0x000c8d, 0x000c8d, PG_U_UNASSIGNED},
{0x000c8e, 0x000c90, PG_U_OTHER_LETTER},
- {0x000c91, 0x000c91, PG_U_UNASSIGNED},
{0x000c92, 0x000ca8, PG_U_OTHER_LETTER},
- {0x000ca9, 0x000ca9, PG_U_UNASSIGNED},
{0x000caa, 0x000cb3, PG_U_OTHER_LETTER},
- {0x000cb4, 0x000cb4, PG_U_UNASSIGNED},
{0x000cb5, 0x000cb9, PG_U_OTHER_LETTER},
- {0x000cba, 0x000cbb, PG_U_UNASSIGNED},
{0x000cbc, 0x000cbc, PG_U_NONSPACING_MARK},
{0x000cbd, 0x000cbd, PG_U_OTHER_LETTER},
{0x000cbe, 0x000cbe, PG_U_SPACING_MARK},
{0x000cbf, 0x000cbf, PG_U_NONSPACING_MARK},
{0x000cc0, 0x000cc4, PG_U_SPACING_MARK},
- {0x000cc5, 0x000cc5, PG_U_UNASSIGNED},
{0x000cc6, 0x000cc6, PG_U_NONSPACING_MARK},
{0x000cc7, 0x000cc8, PG_U_SPACING_MARK},
- {0x000cc9, 0x000cc9, PG_U_UNASSIGNED},
{0x000cca, 0x000ccb, PG_U_SPACING_MARK},
{0x000ccc, 0x000ccd, PG_U_NONSPACING_MARK},
- {0x000cce, 0x000cd4, PG_U_UNASSIGNED},
{0x000cd5, 0x000cd6, PG_U_SPACING_MARK},
- {0x000cd7, 0x000cdc, PG_U_UNASSIGNED},
{0x000cdd, 0x000cde, PG_U_OTHER_LETTER},
- {0x000cdf, 0x000cdf, PG_U_UNASSIGNED},
{0x000ce0, 0x000ce1, PG_U_OTHER_LETTER},
{0x000ce2, 0x000ce3, PG_U_NONSPACING_MARK},
- {0x000ce4, 0x000ce5, PG_U_UNASSIGNED},
{0x000ce6, 0x000cef, PG_U_DECIMAL_NUMBER},
- {0x000cf0, 0x000cf0, PG_U_UNASSIGNED},
{0x000cf1, 0x000cf2, PG_U_OTHER_LETTER},
{0x000cf3, 0x000cf3, PG_U_SPACING_MARK},
- {0x000cf4, 0x000cff, PG_U_UNASSIGNED},
{0x000d00, 0x000d01, PG_U_NONSPACING_MARK},
{0x000d02, 0x000d03, PG_U_SPACING_MARK},
{0x000d04, 0x000d0c, PG_U_OTHER_LETTER},
- {0x000d0d, 0x000d0d, PG_U_UNASSIGNED},
{0x000d0e, 0x000d10, PG_U_OTHER_LETTER},
- {0x000d11, 0x000d11, PG_U_UNASSIGNED},
{0x000d12, 0x000d3a, PG_U_OTHER_LETTER},
{0x000d3b, 0x000d3c, PG_U_NONSPACING_MARK},
{0x000d3d, 0x000d3d, PG_U_OTHER_LETTER},
{0x000d3e, 0x000d40, PG_U_SPACING_MARK},
{0x000d41, 0x000d44, PG_U_NONSPACING_MARK},
- {0x000d45, 0x000d45, PG_U_UNASSIGNED},
{0x000d46, 0x000d48, PG_U_SPACING_MARK},
- {0x000d49, 0x000d49, PG_U_UNASSIGNED},
{0x000d4a, 0x000d4c, PG_U_SPACING_MARK},
{0x000d4d, 0x000d4d, PG_U_NONSPACING_MARK},
{0x000d4e, 0x000d4e, PG_U_OTHER_LETTER},
{0x000d4f, 0x000d4f, PG_U_OTHER_SYMBOL},
- {0x000d50, 0x000d53, PG_U_UNASSIGNED},
{0x000d54, 0x000d56, PG_U_OTHER_LETTER},
{0x000d57, 0x000d57, PG_U_SPACING_MARK},
{0x000d58, 0x000d5e, PG_U_OTHER_NUMBER},
{0x000d5f, 0x000d61, PG_U_OTHER_LETTER},
{0x000d62, 0x000d63, PG_U_NONSPACING_MARK},
- {0x000d64, 0x000d65, PG_U_UNASSIGNED},
{0x000d66, 0x000d6f, PG_U_DECIMAL_NUMBER},
{0x000d70, 0x000d78, PG_U_OTHER_NUMBER},
{0x000d79, 0x000d79, PG_U_OTHER_SYMBOL},
{0x000d7a, 0x000d7f, PG_U_OTHER_LETTER},
- {0x000d80, 0x000d80, PG_U_UNASSIGNED},
{0x000d81, 0x000d81, PG_U_NONSPACING_MARK},
{0x000d82, 0x000d83, PG_U_SPACING_MARK},
- {0x000d84, 0x000d84, PG_U_UNASSIGNED},
{0x000d85, 0x000d96, PG_U_OTHER_LETTER},
- {0x000d97, 0x000d99, PG_U_UNASSIGNED},
{0x000d9a, 0x000db1, PG_U_OTHER_LETTER},
- {0x000db2, 0x000db2, PG_U_UNASSIGNED},
{0x000db3, 0x000dbb, PG_U_OTHER_LETTER},
- {0x000dbc, 0x000dbc, PG_U_UNASSIGNED},
{0x000dbd, 0x000dbd, PG_U_OTHER_LETTER},
- {0x000dbe, 0x000dbf, PG_U_UNASSIGNED},
{0x000dc0, 0x000dc6, PG_U_OTHER_LETTER},
- {0x000dc7, 0x000dc9, PG_U_UNASSIGNED},
{0x000dca, 0x000dca, PG_U_NONSPACING_MARK},
- {0x000dcb, 0x000dce, PG_U_UNASSIGNED},
{0x000dcf, 0x000dd1, PG_U_SPACING_MARK},
{0x000dd2, 0x000dd4, PG_U_NONSPACING_MARK},
- {0x000dd5, 0x000dd5, PG_U_UNASSIGNED},
{0x000dd6, 0x000dd6, PG_U_NONSPACING_MARK},
- {0x000dd7, 0x000dd7, PG_U_UNASSIGNED},
{0x000dd8, 0x000ddf, PG_U_SPACING_MARK},
- {0x000de0, 0x000de5, PG_U_UNASSIGNED},
{0x000de6, 0x000def, PG_U_DECIMAL_NUMBER},
- {0x000df0, 0x000df1, PG_U_UNASSIGNED},
{0x000df2, 0x000df3, PG_U_SPACING_MARK},
{0x000df4, 0x000df4, PG_U_OTHER_PUNCTUATION},
- {0x000df5, 0x000e00, PG_U_UNASSIGNED},
{0x000e01, 0x000e30, PG_U_OTHER_LETTER},
{0x000e31, 0x000e31, PG_U_NONSPACING_MARK},
{0x000e32, 0x000e33, PG_U_OTHER_LETTER},
{0x000e34, 0x000e3a, PG_U_NONSPACING_MARK},
- {0x000e3b, 0x000e3e, PG_U_UNASSIGNED},
{0x000e3f, 0x000e3f, PG_U_CURRENCY_SYMBOL},
{0x000e40, 0x000e45, PG_U_OTHER_LETTER},
{0x000e46, 0x000e46, PG_U_MODIFIER_LETTER},
@@ -1127,33 +985,21 @@ static const pg_category_range unicode_categories[4009] =
{0x000e4f, 0x000e4f, PG_U_OTHER_PUNCTUATION},
{0x000e50, 0x000e59, PG_U_DECIMAL_NUMBER},
{0x000e5a, 0x000e5b, PG_U_OTHER_PUNCTUATION},
- {0x000e5c, 0x000e80, PG_U_UNASSIGNED},
{0x000e81, 0x000e82, PG_U_OTHER_LETTER},
- {0x000e83, 0x000e83, PG_U_UNASSIGNED},
{0x000e84, 0x000e84, PG_U_OTHER_LETTER},
- {0x000e85, 0x000e85, PG_U_UNASSIGNED},
{0x000e86, 0x000e8a, PG_U_OTHER_LETTER},
- {0x000e8b, 0x000e8b, PG_U_UNASSIGNED},
{0x000e8c, 0x000ea3, PG_U_OTHER_LETTER},
- {0x000ea4, 0x000ea4, PG_U_UNASSIGNED},
{0x000ea5, 0x000ea5, PG_U_OTHER_LETTER},
- {0x000ea6, 0x000ea6, PG_U_UNASSIGNED},
{0x000ea7, 0x000eb0, PG_U_OTHER_LETTER},
{0x000eb1, 0x000eb1, PG_U_NONSPACING_MARK},
{0x000eb2, 0x000eb3, PG_U_OTHER_LETTER},
{0x000eb4, 0x000ebc, PG_U_NONSPACING_MARK},
{0x000ebd, 0x000ebd, PG_U_OTHER_LETTER},
- {0x000ebe, 0x000ebf, PG_U_UNASSIGNED},
{0x000ec0, 0x000ec4, PG_U_OTHER_LETTER},
- {0x000ec5, 0x000ec5, PG_U_UNASSIGNED},
{0x000ec6, 0x000ec6, PG_U_MODIFIER_LETTER},
- {0x000ec7, 0x000ec7, PG_U_UNASSIGNED},
{0x000ec8, 0x000ece, PG_U_NONSPACING_MARK},
- {0x000ecf, 0x000ecf, PG_U_UNASSIGNED},
{0x000ed0, 0x000ed9, PG_U_DECIMAL_NUMBER},
- {0x000eda, 0x000edb, PG_U_UNASSIGNED},
{0x000edc, 0x000edf, PG_U_OTHER_LETTER},
- {0x000ee0, 0x000eff, PG_U_UNASSIGNED},
{0x000f00, 0x000f00, PG_U_OTHER_LETTER},
{0x000f01, 0x000f03, PG_U_OTHER_SYMBOL},
{0x000f04, 0x000f12, PG_U_OTHER_PUNCTUATION},
@@ -1176,9 +1022,7 @@ static const pg_category_range unicode_categories[4009] =
{0x000f3d, 0x000f3d, PG_U_CLOSE_PUNCTUATION},
{0x000f3e, 0x000f3f, PG_U_SPACING_MARK},
{0x000f40, 0x000f47, PG_U_OTHER_LETTER},
- {0x000f48, 0x000f48, PG_U_UNASSIGNED},
{0x000f49, 0x000f6c, PG_U_OTHER_LETTER},
- {0x000f6d, 0x000f70, PG_U_UNASSIGNED},
{0x000f71, 0x000f7e, PG_U_NONSPACING_MARK},
{0x000f7f, 0x000f7f, PG_U_SPACING_MARK},
{0x000f80, 0x000f84, PG_U_NONSPACING_MARK},
@@ -1186,18 +1030,14 @@ static const pg_category_range unicode_categories[4009] =
{0x000f86, 0x000f87, PG_U_NONSPACING_MARK},
{0x000f88, 0x000f8c, PG_U_OTHER_LETTER},
{0x000f8d, 0x000f97, PG_U_NONSPACING_MARK},
- {0x000f98, 0x000f98, PG_U_UNASSIGNED},
{0x000f99, 0x000fbc, PG_U_NONSPACING_MARK},
- {0x000fbd, 0x000fbd, PG_U_UNASSIGNED},
{0x000fbe, 0x000fc5, PG_U_OTHER_SYMBOL},
{0x000fc6, 0x000fc6, PG_U_NONSPACING_MARK},
{0x000fc7, 0x000fcc, PG_U_OTHER_SYMBOL},
- {0x000fcd, 0x000fcd, PG_U_UNASSIGNED},
{0x000fce, 0x000fcf, PG_U_OTHER_SYMBOL},
{0x000fd0, 0x000fd4, PG_U_OTHER_PUNCTUATION},
{0x000fd5, 0x000fd8, PG_U_OTHER_SYMBOL},
{0x000fd9, 0x000fda, PG_U_OTHER_PUNCTUATION},
- {0x000fdb, 0x000fff, PG_U_UNASSIGNED},
{0x001000, 0x00102a, PG_U_OTHER_LETTER},
{0x00102b, 0x00102c, PG_U_SPACING_MARK},
{0x00102d, 0x001030, PG_U_NONSPACING_MARK},
@@ -1234,58 +1074,35 @@ static const pg_category_range unicode_categories[4009] =
{0x00109d, 0x00109d, PG_U_NONSPACING_MARK},
{0x00109e, 0x00109f, PG_U_OTHER_SYMBOL},
{0x0010a0, 0x0010c5, PG_U_UPPERCASE_LETTER},
- {0x0010c6, 0x0010c6, PG_U_UNASSIGNED},
{0x0010c7, 0x0010c7, PG_U_UPPERCASE_LETTER},
- {0x0010c8, 0x0010cc, PG_U_UNASSIGNED},
{0x0010cd, 0x0010cd, PG_U_UPPERCASE_LETTER},
- {0x0010ce, 0x0010cf, PG_U_UNASSIGNED},
{0x0010d0, 0x0010fa, PG_U_LOWERCASE_LETTER},
{0x0010fb, 0x0010fb, PG_U_OTHER_PUNCTUATION},
{0x0010fc, 0x0010fc, PG_U_MODIFIER_LETTER},
{0x0010fd, 0x0010ff, PG_U_LOWERCASE_LETTER},
{0x001100, 0x001248, PG_U_OTHER_LETTER},
- {0x001249, 0x001249, PG_U_UNASSIGNED},
{0x00124a, 0x00124d, PG_U_OTHER_LETTER},
- {0x00124e, 0x00124f, PG_U_UNASSIGNED},
{0x001250, 0x001256, PG_U_OTHER_LETTER},
- {0x001257, 0x001257, PG_U_UNASSIGNED},
{0x001258, 0x001258, PG_U_OTHER_LETTER},
- {0x001259, 0x001259, PG_U_UNASSIGNED},
{0x00125a, 0x00125d, PG_U_OTHER_LETTER},
- {0x00125e, 0x00125f, PG_U_UNASSIGNED},
{0x001260, 0x001288, PG_U_OTHER_LETTER},
- {0x001289, 0x001289, PG_U_UNASSIGNED},
{0x00128a, 0x00128d, PG_U_OTHER_LETTER},
- {0x00128e, 0x00128f, PG_U_UNASSIGNED},
{0x001290, 0x0012b0, PG_U_OTHER_LETTER},
- {0x0012b1, 0x0012b1, PG_U_UNASSIGNED},
{0x0012b2, 0x0012b5, PG_U_OTHER_LETTER},
- {0x0012b6, 0x0012b7, PG_U_UNASSIGNED},
{0x0012b8, 0x0012be, PG_U_OTHER_LETTER},
- {0x0012bf, 0x0012bf, PG_U_UNASSIGNED},
{0x0012c0, 0x0012c0, PG_U_OTHER_LETTER},
- {0x0012c1, 0x0012c1, PG_U_UNASSIGNED},
{0x0012c2, 0x0012c5, PG_U_OTHER_LETTER},
- {0x0012c6, 0x0012c7, PG_U_UNASSIGNED},
{0x0012c8, 0x0012d6, PG_U_OTHER_LETTER},
- {0x0012d7, 0x0012d7, PG_U_UNASSIGNED},
{0x0012d8, 0x001310, PG_U_OTHER_LETTER},
- {0x001311, 0x001311, PG_U_UNASSIGNED},
{0x001312, 0x001315, PG_U_OTHER_LETTER},
- {0x001316, 0x001317, PG_U_UNASSIGNED},
{0x001318, 0x00135a, PG_U_OTHER_LETTER},
- {0x00135b, 0x00135c, PG_U_UNASSIGNED},
{0x00135d, 0x00135f, PG_U_NONSPACING_MARK},
{0x001360, 0x001368, PG_U_OTHER_PUNCTUATION},
{0x001369, 0x00137c, PG_U_OTHER_NUMBER},
- {0x00137d, 0x00137f, PG_U_UNASSIGNED},
{0x001380, 0x00138f, PG_U_OTHER_LETTER},
{0x001390, 0x001399, PG_U_OTHER_SYMBOL},
- {0x00139a, 0x00139f, PG_U_UNASSIGNED},
{0x0013a0, 0x0013f5, PG_U_UPPERCASE_LETTER},
- {0x0013f6, 0x0013f7, PG_U_UNASSIGNED},
{0x0013f8, 0x0013fd, PG_U_LOWERCASE_LETTER},
- {0x0013fe, 0x0013ff, PG_U_UNASSIGNED},
{0x001400, 0x001400, PG_U_DASH_PUNCTUATION},
{0x001401, 0x00166c, PG_U_OTHER_LETTER},
{0x00166d, 0x00166d, PG_U_OTHER_SYMBOL},
@@ -1295,30 +1112,22 @@ static const pg_category_range unicode_categories[4009] =
{0x001681, 0x00169a, PG_U_OTHER_LETTER},
{0x00169b, 0x00169b, PG_U_OPEN_PUNCTUATION},
{0x00169c, 0x00169c, PG_U_CLOSE_PUNCTUATION},
- {0x00169d, 0x00169f, PG_U_UNASSIGNED},
{0x0016a0, 0x0016ea, PG_U_OTHER_LETTER},
{0x0016eb, 0x0016ed, PG_U_OTHER_PUNCTUATION},
{0x0016ee, 0x0016f0, PG_U_LETTER_NUMBER},
{0x0016f1, 0x0016f8, PG_U_OTHER_LETTER},
- {0x0016f9, 0x0016ff, PG_U_UNASSIGNED},
{0x001700, 0x001711, PG_U_OTHER_LETTER},
{0x001712, 0x001714, PG_U_NONSPACING_MARK},
{0x001715, 0x001715, PG_U_SPACING_MARK},
- {0x001716, 0x00171e, PG_U_UNASSIGNED},
{0x00171f, 0x001731, PG_U_OTHER_LETTER},
{0x001732, 0x001733, PG_U_NONSPACING_MARK},
{0x001734, 0x001734, PG_U_SPACING_MARK},
{0x001735, 0x001736, PG_U_OTHER_PUNCTUATION},
- {0x001737, 0x00173f, PG_U_UNASSIGNED},
{0x001740, 0x001751, PG_U_OTHER_LETTER},
{0x001752, 0x001753, PG_U_NONSPACING_MARK},
- {0x001754, 0x00175f, PG_U_UNASSIGNED},
{0x001760, 0x00176c, PG_U_OTHER_LETTER},
- {0x00176d, 0x00176d, PG_U_UNASSIGNED},
{0x00176e, 0x001770, PG_U_OTHER_LETTER},
- {0x001771, 0x001771, PG_U_UNASSIGNED},
{0x001772, 0x001773, PG_U_NONSPACING_MARK},
- {0x001774, 0x00177f, PG_U_UNASSIGNED},
{0x001780, 0x0017b3, PG_U_OTHER_LETTER},
{0x0017b4, 0x0017b5, PG_U_NONSPACING_MARK},
{0x0017b6, 0x0017b6, PG_U_SPACING_MARK},
@@ -1333,11 +1142,8 @@ static const pg_category_range unicode_categories[4009] =
{0x0017db, 0x0017db, PG_U_CURRENCY_SYMBOL},
{0x0017dc, 0x0017dc, PG_U_OTHER_LETTER},
{0x0017dd, 0x0017dd, PG_U_NONSPACING_MARK},
- {0x0017de, 0x0017df, PG_U_UNASSIGNED},
{0x0017e0, 0x0017e9, PG_U_DECIMAL_NUMBER},
- {0x0017ea, 0x0017ef, PG_U_UNASSIGNED},
{0x0017f0, 0x0017f9, PG_U_OTHER_NUMBER},
- {0x0017fa, 0x0017ff, PG_U_UNASSIGNED},
{0x001800, 0x001805, PG_U_OTHER_PUNCTUATION},
{0x001806, 0x001806, PG_U_DASH_PUNCTUATION},
{0x001807, 0x00180a, PG_U_OTHER_PUNCTUATION},
@@ -1345,59 +1151,44 @@ static const pg_category_range unicode_categories[4009] =
{0x00180e, 0x00180e, PG_U_FORMAT},
{0x00180f, 0x00180f, PG_U_NONSPACING_MARK},
{0x001810, 0x001819, PG_U_DECIMAL_NUMBER},
- {0x00181a, 0x00181f, PG_U_UNASSIGNED},
{0x001820, 0x001842, PG_U_OTHER_LETTER},
{0x001843, 0x001843, PG_U_MODIFIER_LETTER},
{0x001844, 0x001878, PG_U_OTHER_LETTER},
- {0x001879, 0x00187f, PG_U_UNASSIGNED},
{0x001880, 0x001884, PG_U_OTHER_LETTER},
{0x001885, 0x001886, PG_U_NONSPACING_MARK},
{0x001887, 0x0018a8, PG_U_OTHER_LETTER},
{0x0018a9, 0x0018a9, PG_U_NONSPACING_MARK},
{0x0018aa, 0x0018aa, PG_U_OTHER_LETTER},
- {0x0018ab, 0x0018af, PG_U_UNASSIGNED},
{0x0018b0, 0x0018f5, PG_U_OTHER_LETTER},
- {0x0018f6, 0x0018ff, PG_U_UNASSIGNED},
{0x001900, 0x00191e, PG_U_OTHER_LETTER},
- {0x00191f, 0x00191f, PG_U_UNASSIGNED},
{0x001920, 0x001922, PG_U_NONSPACING_MARK},
{0x001923, 0x001926, PG_U_SPACING_MARK},
{0x001927, 0x001928, PG_U_NONSPACING_MARK},
{0x001929, 0x00192b, PG_U_SPACING_MARK},
- {0x00192c, 0x00192f, PG_U_UNASSIGNED},
{0x001930, 0x001931, PG_U_SPACING_MARK},
{0x001932, 0x001932, PG_U_NONSPACING_MARK},
{0x001933, 0x001938, PG_U_SPACING_MARK},
{0x001939, 0x00193b, PG_U_NONSPACING_MARK},
- {0x00193c, 0x00193f, PG_U_UNASSIGNED},
{0x001940, 0x001940, PG_U_OTHER_SYMBOL},
- {0x001941, 0x001943, PG_U_UNASSIGNED},
{0x001944, 0x001945, PG_U_OTHER_PUNCTUATION},
{0x001946, 0x00194f, PG_U_DECIMAL_NUMBER},
{0x001950, 0x00196d, PG_U_OTHER_LETTER},
- {0x00196e, 0x00196f, PG_U_UNASSIGNED},
{0x001970, 0x001974, PG_U_OTHER_LETTER},
- {0x001975, 0x00197f, PG_U_UNASSIGNED},
{0x001980, 0x0019ab, PG_U_OTHER_LETTER},
- {0x0019ac, 0x0019af, PG_U_UNASSIGNED},
{0x0019b0, 0x0019c9, PG_U_OTHER_LETTER},
- {0x0019ca, 0x0019cf, PG_U_UNASSIGNED},
{0x0019d0, 0x0019d9, PG_U_DECIMAL_NUMBER},
{0x0019da, 0x0019da, PG_U_OTHER_NUMBER},
- {0x0019db, 0x0019dd, PG_U_UNASSIGNED},
{0x0019de, 0x0019ff, PG_U_OTHER_SYMBOL},
{0x001a00, 0x001a16, PG_U_OTHER_LETTER},
{0x001a17, 0x001a18, PG_U_NONSPACING_MARK},
{0x001a19, 0x001a1a, PG_U_SPACING_MARK},
{0x001a1b, 0x001a1b, PG_U_NONSPACING_MARK},
- {0x001a1c, 0x001a1d, PG_U_UNASSIGNED},
{0x001a1e, 0x001a1f, PG_U_OTHER_PUNCTUATION},
{0x001a20, 0x001a54, PG_U_OTHER_LETTER},
{0x001a55, 0x001a55, PG_U_SPACING_MARK},
{0x001a56, 0x001a56, PG_U_NONSPACING_MARK},
{0x001a57, 0x001a57, PG_U_SPACING_MARK},
{0x001a58, 0x001a5e, PG_U_NONSPACING_MARK},
- {0x001a5f, 0x001a5f, PG_U_UNASSIGNED},
{0x001a60, 0x001a60, PG_U_NONSPACING_MARK},
{0x001a61, 0x001a61, PG_U_SPACING_MARK},
{0x001a62, 0x001a62, PG_U_NONSPACING_MARK},
@@ -1405,20 +1196,15 @@ static const pg_category_range unicode_categories[4009] =
{0x001a65, 0x001a6c, PG_U_NONSPACING_MARK},
{0x001a6d, 0x001a72, PG_U_SPACING_MARK},
{0x001a73, 0x001a7c, PG_U_NONSPACING_MARK},
- {0x001a7d, 0x001a7e, PG_U_UNASSIGNED},
{0x001a7f, 0x001a7f, PG_U_NONSPACING_MARK},
{0x001a80, 0x001a89, PG_U_DECIMAL_NUMBER},
- {0x001a8a, 0x001a8f, PG_U_UNASSIGNED},
{0x001a90, 0x001a99, PG_U_DECIMAL_NUMBER},
- {0x001a9a, 0x001a9f, PG_U_UNASSIGNED},
{0x001aa0, 0x001aa6, PG_U_OTHER_PUNCTUATION},
{0x001aa7, 0x001aa7, PG_U_MODIFIER_LETTER},
{0x001aa8, 0x001aad, PG_U_OTHER_PUNCTUATION},
- {0x001aae, 0x001aaf, PG_U_UNASSIGNED},
{0x001ab0, 0x001abd, PG_U_NONSPACING_MARK},
{0x001abe, 0x001abe, PG_U_ENCLOSING_MARK},
{0x001abf, 0x001ace, PG_U_NONSPACING_MARK},
- {0x001acf, 0x001aff, PG_U_UNASSIGNED},
{0x001b00, 0x001b03, PG_U_NONSPACING_MARK},
{0x001b04, 0x001b04, PG_U_SPACING_MARK},
{0x001b05, 0x001b33, PG_U_OTHER_LETTER},
@@ -1431,14 +1217,12 @@ static const pg_category_range unicode_categories[4009] =
{0x001b42, 0x001b42, PG_U_NONSPACING_MARK},
{0x001b43, 0x001b44, PG_U_SPACING_MARK},
{0x001b45, 0x001b4c, PG_U_OTHER_LETTER},
- {0x001b4d, 0x001b4f, PG_U_UNASSIGNED},
{0x001b50, 0x001b59, PG_U_DECIMAL_NUMBER},
{0x001b5a, 0x001b60, PG_U_OTHER_PUNCTUATION},
{0x001b61, 0x001b6a, PG_U_OTHER_SYMBOL},
{0x001b6b, 0x001b73, PG_U_NONSPACING_MARK},
{0x001b74, 0x001b7c, PG_U_OTHER_SYMBOL},
{0x001b7d, 0x001b7e, PG_U_OTHER_PUNCTUATION},
- {0x001b7f, 0x001b7f, PG_U_UNASSIGNED},
{0x001b80, 0x001b81, PG_U_NONSPACING_MARK},
{0x001b82, 0x001b82, PG_U_SPACING_MARK},
{0x001b83, 0x001ba0, PG_U_OTHER_LETTER},
@@ -1459,29 +1243,23 @@ static const pg_category_range unicode_categories[4009] =
{0x001bee, 0x001bee, PG_U_SPACING_MARK},
{0x001bef, 0x001bf1, PG_U_NONSPACING_MARK},
{0x001bf2, 0x001bf3, PG_U_SPACING_MARK},
- {0x001bf4, 0x001bfb, PG_U_UNASSIGNED},
{0x001bfc, 0x001bff, PG_U_OTHER_PUNCTUATION},
{0x001c00, 0x001c23, PG_U_OTHER_LETTER},
{0x001c24, 0x001c2b, PG_U_SPACING_MARK},
{0x001c2c, 0x001c33, PG_U_NONSPACING_MARK},
{0x001c34, 0x001c35, PG_U_SPACING_MARK},
{0x001c36, 0x001c37, PG_U_NONSPACING_MARK},
- {0x001c38, 0x001c3a, PG_U_UNASSIGNED},
{0x001c3b, 0x001c3f, PG_U_OTHER_PUNCTUATION},
{0x001c40, 0x001c49, PG_U_DECIMAL_NUMBER},
- {0x001c4a, 0x001c4c, PG_U_UNASSIGNED},
{0x001c4d, 0x001c4f, PG_U_OTHER_LETTER},
{0x001c50, 0x001c59, PG_U_DECIMAL_NUMBER},
{0x001c5a, 0x001c77, PG_U_OTHER_LETTER},
{0x001c78, 0x001c7d, PG_U_MODIFIER_LETTER},
{0x001c7e, 0x001c7f, PG_U_OTHER_PUNCTUATION},
{0x001c80, 0x001c88, PG_U_LOWERCASE_LETTER},
- {0x001c89, 0x001c8f, PG_U_UNASSIGNED},
{0x001c90, 0x001cba, PG_U_UPPERCASE_LETTER},
- {0x001cbb, 0x001cbc, PG_U_UNASSIGNED},
{0x001cbd, 0x001cbf, PG_U_UPPERCASE_LETTER},
{0x001cc0, 0x001cc7, PG_U_OTHER_PUNCTUATION},
- {0x001cc8, 0x001ccf, PG_U_UNASSIGNED},
{0x001cd0, 0x001cd2, PG_U_NONSPACING_MARK},
{0x001cd3, 0x001cd3, PG_U_OTHER_PUNCTUATION},
{0x001cd4, 0x001ce0, PG_U_NONSPACING_MARK},
@@ -1495,7 +1273,6 @@ static const pg_category_range unicode_categories[4009] =
{0x001cf7, 0x001cf7, PG_U_SPACING_MARK},
{0x001cf8, 0x001cf9, PG_U_NONSPACING_MARK},
{0x001cfa, 0x001cfa, PG_U_OTHER_LETTER},
- {0x001cfb, 0x001cff, PG_U_UNASSIGNED},
{0x001d00, 0x001d2b, PG_U_LOWERCASE_LETTER},
{0x001d2c, 0x001d6a, PG_U_MODIFIER_LETTER},
{0x001d6b, 0x001d77, PG_U_LOWERCASE_LETTER},
@@ -1753,30 +1530,21 @@ static const pg_category_range unicode_categories[4009] =
{0x001eff, 0x001f07, PG_U_LOWERCASE_LETTER},
{0x001f08, 0x001f0f, PG_U_UPPERCASE_LETTER},
{0x001f10, 0x001f15, PG_U_LOWERCASE_LETTER},
- {0x001f16, 0x001f17, PG_U_UNASSIGNED},
{0x001f18, 0x001f1d, PG_U_UPPERCASE_LETTER},
- {0x001f1e, 0x001f1f, PG_U_UNASSIGNED},
{0x001f20, 0x001f27, PG_U_LOWERCASE_LETTER},
{0x001f28, 0x001f2f, PG_U_UPPERCASE_LETTER},
{0x001f30, 0x001f37, PG_U_LOWERCASE_LETTER},
{0x001f38, 0x001f3f, PG_U_UPPERCASE_LETTER},
{0x001f40, 0x001f45, PG_U_LOWERCASE_LETTER},
- {0x001f46, 0x001f47, PG_U_UNASSIGNED},
{0x001f48, 0x001f4d, PG_U_UPPERCASE_LETTER},
- {0x001f4e, 0x001f4f, PG_U_UNASSIGNED},
{0x001f50, 0x001f57, PG_U_LOWERCASE_LETTER},
- {0x001f58, 0x001f58, PG_U_UNASSIGNED},
{0x001f59, 0x001f59, PG_U_UPPERCASE_LETTER},
- {0x001f5a, 0x001f5a, PG_U_UNASSIGNED},
{0x001f5b, 0x001f5b, PG_U_UPPERCASE_LETTER},
- {0x001f5c, 0x001f5c, PG_U_UNASSIGNED},
{0x001f5d, 0x001f5d, PG_U_UPPERCASE_LETTER},
- {0x001f5e, 0x001f5e, PG_U_UNASSIGNED},
{0x001f5f, 0x001f5f, PG_U_UPPERCASE_LETTER},
{0x001f60, 0x001f67, PG_U_LOWERCASE_LETTER},
{0x001f68, 0x001f6f, PG_U_UPPERCASE_LETTER},
{0x001f70, 0x001f7d, PG_U_LOWERCASE_LETTER},
- {0x001f7e, 0x001f7f, PG_U_UNASSIGNED},
{0x001f80, 0x001f87, PG_U_LOWERCASE_LETTER},
{0x001f88, 0x001f8f, PG_U_TITLECASE_LETTER},
{0x001f90, 0x001f97, PG_U_LOWERCASE_LETTER},
@@ -1784,7 +1552,6 @@ static const pg_category_range unicode_categories[4009] =
{0x001fa0, 0x001fa7, PG_U_LOWERCASE_LETTER},
{0x001fa8, 0x001faf, PG_U_TITLECASE_LETTER},
{0x001fb0, 0x001fb4, PG_U_LOWERCASE_LETTER},
- {0x001fb5, 0x001fb5, PG_U_UNASSIGNED},
{0x001fb6, 0x001fb7, PG_U_LOWERCASE_LETTER},
{0x001fb8, 0x001fbb, PG_U_UPPERCASE_LETTER},
{0x001fbc, 0x001fbc, PG_U_TITLECASE_LETTER},
@@ -1792,28 +1559,22 @@ static const pg_category_range unicode_categories[4009] =
{0x001fbe, 0x001fbe, PG_U_LOWERCASE_LETTER},
{0x001fbf, 0x001fc1, PG_U_MODIFIER_SYMBOL},
{0x001fc2, 0x001fc4, PG_U_LOWERCASE_LETTER},
- {0x001fc5, 0x001fc5, PG_U_UNASSIGNED},
{0x001fc6, 0x001fc7, PG_U_LOWERCASE_LETTER},
{0x001fc8, 0x001fcb, PG_U_UPPERCASE_LETTER},
{0x001fcc, 0x001fcc, PG_U_TITLECASE_LETTER},
{0x001fcd, 0x001fcf, PG_U_MODIFIER_SYMBOL},
{0x001fd0, 0x001fd3, PG_U_LOWERCASE_LETTER},
- {0x001fd4, 0x001fd5, PG_U_UNASSIGNED},
{0x001fd6, 0x001fd7, PG_U_LOWERCASE_LETTER},
{0x001fd8, 0x001fdb, PG_U_UPPERCASE_LETTER},
- {0x001fdc, 0x001fdc, PG_U_UNASSIGNED},
{0x001fdd, 0x001fdf, PG_U_MODIFIER_SYMBOL},
{0x001fe0, 0x001fe7, PG_U_LOWERCASE_LETTER},
{0x001fe8, 0x001fec, PG_U_UPPERCASE_LETTER},
{0x001fed, 0x001fef, PG_U_MODIFIER_SYMBOL},
- {0x001ff0, 0x001ff1, PG_U_UNASSIGNED},
{0x001ff2, 0x001ff4, PG_U_LOWERCASE_LETTER},
- {0x001ff5, 0x001ff5, PG_U_UNASSIGNED},
{0x001ff6, 0x001ff7, PG_U_LOWERCASE_LETTER},
{0x001ff8, 0x001ffb, PG_U_UPPERCASE_LETTER},
{0x001ffc, 0x001ffc, PG_U_TITLECASE_LETTER},
{0x001ffd, 0x001ffe, PG_U_MODIFIER_SYMBOL},
- {0x001fff, 0x001fff, PG_U_UNASSIGNED},
{0x002000, 0x00200a, PG_U_SPACE_SEPARATOR},
{0x00200b, 0x00200f, PG_U_FORMAT},
{0x002010, 0x002015, PG_U_DASH_PUNCTUATION},
@@ -1846,11 +1607,9 @@ static const pg_category_range unicode_categories[4009] =
{0x002055, 0x00205e, PG_U_OTHER_PUNCTUATION},
{0x00205f, 0x00205f, PG_U_SPACE_SEPARATOR},
{0x002060, 0x002064, PG_U_FORMAT},
- {0x002065, 0x002065, PG_U_UNASSIGNED},
{0x002066, 0x00206f, PG_U_FORMAT},
{0x002070, 0x002070, PG_U_OTHER_NUMBER},
{0x002071, 0x002071, PG_U_MODIFIER_LETTER},
- {0x002072, 0x002073, PG_U_UNASSIGNED},
{0x002074, 0x002079, PG_U_OTHER_NUMBER},
{0x00207a, 0x00207c, PG_U_MATH_SYMBOL},
{0x00207d, 0x00207d, PG_U_OPEN_PUNCTUATION},
@@ -1860,17 +1619,13 @@ static const pg_category_range unicode_categories[4009] =
{0x00208a, 0x00208c, PG_U_MATH_SYMBOL},
{0x00208d, 0x00208d, PG_U_OPEN_PUNCTUATION},
{0x00208e, 0x00208e, PG_U_CLOSE_PUNCTUATION},
- {0x00208f, 0x00208f, PG_U_UNASSIGNED},
{0x002090, 0x00209c, PG_U_MODIFIER_LETTER},
- {0x00209d, 0x00209f, PG_U_UNASSIGNED},
{0x0020a0, 0x0020c0, PG_U_CURRENCY_SYMBOL},
- {0x0020c1, 0x0020cf, PG_U_UNASSIGNED},
{0x0020d0, 0x0020dc, PG_U_NONSPACING_MARK},
{0x0020dd, 0x0020e0, PG_U_ENCLOSING_MARK},
{0x0020e1, 0x0020e1, PG_U_NONSPACING_MARK},
{0x0020e2, 0x0020e4, PG_U_ENCLOSING_MARK},
{0x0020e5, 0x0020f0, PG_U_NONSPACING_MARK},
- {0x0020f1, 0x0020ff, PG_U_UNASSIGNED},
{0x002100, 0x002101, PG_U_OTHER_SYMBOL},
{0x002102, 0x002102, PG_U_UPPERCASE_LETTER},
{0x002103, 0x002106, PG_U_OTHER_SYMBOL},
@@ -1918,7 +1673,6 @@ static const pg_category_range unicode_categories[4009] =
{0x002185, 0x002188, PG_U_LETTER_NUMBER},
{0x002189, 0x002189, PG_U_OTHER_NUMBER},
{0x00218a, 0x00218b, PG_U_OTHER_SYMBOL},
- {0x00218c, 0x00218f, PG_U_UNASSIGNED},
{0x002190, 0x002194, PG_U_MATH_SYMBOL},
{0x002195, 0x002199, PG_U_OTHER_SYMBOL},
{0x00219a, 0x00219b, PG_U_MATH_SYMBOL},
@@ -1955,9 +1709,7 @@ static const pg_category_range unicode_categories[4009] =
{0x0023b4, 0x0023db, PG_U_OTHER_SYMBOL},
{0x0023dc, 0x0023e1, PG_U_MATH_SYMBOL},
{0x0023e2, 0x002426, PG_U_OTHER_SYMBOL},
- {0x002427, 0x00243f, PG_U_UNASSIGNED},
{0x002440, 0x00244a, PG_U_OTHER_SYMBOL},
- {0x00244b, 0x00245f, PG_U_UNASSIGNED},
{0x002460, 0x00249b, PG_U_OTHER_NUMBER},
{0x00249c, 0x0024e9, PG_U_OTHER_SYMBOL},
{0x0024ea, 0x0024ff, PG_U_OTHER_NUMBER},
@@ -2039,9 +1791,7 @@ static const pg_category_range unicode_categories[4009] =
{0x002b45, 0x002b46, PG_U_OTHER_SYMBOL},
{0x002b47, 0x002b4c, PG_U_MATH_SYMBOL},
{0x002b4d, 0x002b73, PG_U_OTHER_SYMBOL},
- {0x002b74, 0x002b75, PG_U_UNASSIGNED},
{0x002b76, 0x002b95, PG_U_OTHER_SYMBOL},
- {0x002b96, 0x002b96, PG_U_UNASSIGNED},
{0x002b97, 0x002bff, PG_U_OTHER_SYMBOL},
{0x002c00, 0x002c2f, PG_U_UPPERCASE_LETTER},
{0x002c30, 0x002c5f, PG_U_LOWERCASE_LETTER},
@@ -2170,40 +1920,25 @@ static const pg_category_range unicode_categories[4009] =
{0x002cef, 0x002cf1, PG_U_NONSPACING_MARK},
{0x002cf2, 0x002cf2, PG_U_UPPERCASE_LETTER},
{0x002cf3, 0x002cf3, PG_U_LOWERCASE_LETTER},
- {0x002cf4, 0x002cf8, PG_U_UNASSIGNED},
{0x002cf9, 0x002cfc, PG_U_OTHER_PUNCTUATION},
{0x002cfd, 0x002cfd, PG_U_OTHER_NUMBER},
{0x002cfe, 0x002cff, PG_U_OTHER_PUNCTUATION},
{0x002d00, 0x002d25, PG_U_LOWERCASE_LETTER},
- {0x002d26, 0x002d26, PG_U_UNASSIGNED},
{0x002d27, 0x002d27, PG_U_LOWERCASE_LETTER},
- {0x002d28, 0x002d2c, PG_U_UNASSIGNED},
{0x002d2d, 0x002d2d, PG_U_LOWERCASE_LETTER},
- {0x002d2e, 0x002d2f, PG_U_UNASSIGNED},
{0x002d30, 0x002d67, PG_U_OTHER_LETTER},
- {0x002d68, 0x002d6e, PG_U_UNASSIGNED},
{0x002d6f, 0x002d6f, PG_U_MODIFIER_LETTER},
{0x002d70, 0x002d70, PG_U_OTHER_PUNCTUATION},
- {0x002d71, 0x002d7e, PG_U_UNASSIGNED},
{0x002d7f, 0x002d7f, PG_U_NONSPACING_MARK},
{0x002d80, 0x002d96, PG_U_OTHER_LETTER},
- {0x002d97, 0x002d9f, PG_U_UNASSIGNED},
{0x002da0, 0x002da6, PG_U_OTHER_LETTER},
- {0x002da7, 0x002da7, PG_U_UNASSIGNED},
{0x002da8, 0x002dae, PG_U_OTHER_LETTER},
- {0x002daf, 0x002daf, PG_U_UNASSIGNED},
{0x002db0, 0x002db6, PG_U_OTHER_LETTER},
- {0x002db7, 0x002db7, PG_U_UNASSIGNED},
{0x002db8, 0x002dbe, PG_U_OTHER_LETTER},
- {0x002dbf, 0x002dbf, PG_U_UNASSIGNED},
{0x002dc0, 0x002dc6, PG_U_OTHER_LETTER},
- {0x002dc7, 0x002dc7, PG_U_UNASSIGNED},
{0x002dc8, 0x002dce, PG_U_OTHER_LETTER},
- {0x002dcf, 0x002dcf, PG_U_UNASSIGNED},
{0x002dd0, 0x002dd6, PG_U_OTHER_LETTER},
- {0x002dd7, 0x002dd7, PG_U_UNASSIGNED},
{0x002dd8, 0x002dde, PG_U_OTHER_LETTER},
- {0x002ddf, 0x002ddf, PG_U_UNASSIGNED},
{0x002de0, 0x002dff, PG_U_NONSPACING_MARK},
{0x002e00, 0x002e01, PG_U_OTHER_PUNCTUATION},
{0x002e02, 0x002e02, PG_U_INITIAL_PUNCTUATION},
@@ -2254,13 +1989,9 @@ static const pg_category_range unicode_categories[4009] =
{0x002e5b, 0x002e5b, PG_U_OPEN_PUNCTUATION},
{0x002e5c, 0x002e5c, PG_U_CLOSE_PUNCTUATION},
{0x002e5d, 0x002e5d, PG_U_DASH_PUNCTUATION},
- {0x002e5e, 0x002e7f, PG_U_UNASSIGNED},
{0x002e80, 0x002e99, PG_U_OTHER_SYMBOL},
- {0x002e9a, 0x002e9a, PG_U_UNASSIGNED},
{0x002e9b, 0x002ef3, PG_U_OTHER_SYMBOL},
- {0x002ef4, 0x002eff, PG_U_UNASSIGNED},
{0x002f00, 0x002fd5, PG_U_OTHER_SYMBOL},
- {0x002fd6, 0x002fef, PG_U_UNASSIGNED},
{0x002ff0, 0x002fff, PG_U_OTHER_SYMBOL},
{0x003000, 0x003000, PG_U_SPACE_SEPARATOR},
{0x003001, 0x003003, PG_U_OTHER_PUNCTUATION},
@@ -2302,9 +2033,7 @@ static const pg_category_range unicode_categories[4009] =
{0x00303c, 0x00303c, PG_U_OTHER_LETTER},
{0x00303d, 0x00303d, PG_U_OTHER_PUNCTUATION},
{0x00303e, 0x00303f, PG_U_OTHER_SYMBOL},
- {0x003040, 0x003040, PG_U_UNASSIGNED},
{0x003041, 0x003096, PG_U_OTHER_LETTER},
- {0x003097, 0x003098, PG_U_UNASSIGNED},
{0x003099, 0x00309a, PG_U_NONSPACING_MARK},
{0x00309b, 0x00309c, PG_U_MODIFIER_SYMBOL},
{0x00309d, 0x00309e, PG_U_MODIFIER_LETTER},
@@ -2314,21 +2043,16 @@ static const pg_category_range unicode_categories[4009] =
{0x0030fb, 0x0030fb, PG_U_OTHER_PUNCTUATION},
{0x0030fc, 0x0030fe, PG_U_MODIFIER_LETTER},
{0x0030ff, 0x0030ff, PG_U_OTHER_LETTER},
- {0x003100, 0x003104, PG_U_UNASSIGNED},
{0x003105, 0x00312f, PG_U_OTHER_LETTER},
- {0x003130, 0x003130, PG_U_UNASSIGNED},
{0x003131, 0x00318e, PG_U_OTHER_LETTER},
- {0x00318f, 0x00318f, PG_U_UNASSIGNED},
{0x003190, 0x003191, PG_U_OTHER_SYMBOL},
{0x003192, 0x003195, PG_U_OTHER_NUMBER},
{0x003196, 0x00319f, PG_U_OTHER_SYMBOL},
{0x0031a0, 0x0031bf, PG_U_OTHER_LETTER},
{0x0031c0, 0x0031e3, PG_U_OTHER_SYMBOL},
- {0x0031e4, 0x0031ee, PG_U_UNASSIGNED},
{0x0031ef, 0x0031ef, PG_U_OTHER_SYMBOL},
{0x0031f0, 0x0031ff, PG_U_OTHER_LETTER},
{0x003200, 0x00321e, PG_U_OTHER_SYMBOL},
- {0x00321f, 0x00321f, PG_U_UNASSIGNED},
{0x003220, 0x003229, PG_U_OTHER_NUMBER},
{0x00322a, 0x003247, PG_U_OTHER_SYMBOL},
{0x003248, 0x00324f, PG_U_OTHER_NUMBER},
@@ -2344,9 +2068,7 @@ static const pg_category_range unicode_categories[4009] =
{0x004e00, 0x00a014, PG_U_OTHER_LETTER},
{0x00a015, 0x00a015, PG_U_MODIFIER_LETTER},
{0x00a016, 0x00a48c, PG_U_OTHER_LETTER},
- {0x00a48d, 0x00a48f, PG_U_UNASSIGNED},
{0x00a490, 0x00a4c6, PG_U_OTHER_SYMBOL},
- {0x00a4c7, 0x00a4cf, PG_U_UNASSIGNED},
{0x00a4d0, 0x00a4f7, PG_U_OTHER_LETTER},
{0x00a4f8, 0x00a4fd, PG_U_MODIFIER_LETTER},
{0x00a4fe, 0x00a4ff, PG_U_OTHER_PUNCTUATION},
@@ -2356,7 +2078,6 @@ static const pg_category_range unicode_categories[4009] =
{0x00a610, 0x00a61f, PG_U_OTHER_LETTER},
{0x00a620, 0x00a629, PG_U_DECIMAL_NUMBER},
{0x00a62a, 0x00a62b, PG_U_OTHER_LETTER},
- {0x00a62c, 0x00a63f, PG_U_UNASSIGNED},
{0x00a640, 0x00a640, PG_U_UPPERCASE_LETTER},
{0x00a641, 0x00a641, PG_U_LOWERCASE_LETTER},
{0x00a642, 0x00a642, PG_U_UPPERCASE_LETTER},
@@ -2444,7 +2165,6 @@ static const pg_category_range unicode_categories[4009] =
{0x00a6e6, 0x00a6ef, PG_U_LETTER_NUMBER},
{0x00a6f0, 0x00a6f1, PG_U_NONSPACING_MARK},
{0x00a6f2, 0x00a6f7, PG_U_OTHER_PUNCTUATION},
- {0x00a6f8, 0x00a6ff, PG_U_UNASSIGNED},
{0x00a700, 0x00a716, PG_U_MODIFIER_SYMBOL},
{0x00a717, 0x00a71f, PG_U_MODIFIER_LETTER},
{0x00a720, 0x00a721, PG_U_MODIFIER_SYMBOL},
@@ -2593,18 +2313,14 @@ static const pg_category_range unicode_categories[4009] =
{0x00a7c8, 0x00a7c8, PG_U_LOWERCASE_LETTER},
{0x00a7c9, 0x00a7c9, PG_U_UPPERCASE_LETTER},
{0x00a7ca, 0x00a7ca, PG_U_LOWERCASE_LETTER},
- {0x00a7cb, 0x00a7cf, PG_U_UNASSIGNED},
{0x00a7d0, 0x00a7d0, PG_U_UPPERCASE_LETTER},
{0x00a7d1, 0x00a7d1, PG_U_LOWERCASE_LETTER},
- {0x00a7d2, 0x00a7d2, PG_U_UNASSIGNED},
{0x00a7d3, 0x00a7d3, PG_U_LOWERCASE_LETTER},
- {0x00a7d4, 0x00a7d4, PG_U_UNASSIGNED},
{0x00a7d5, 0x00a7d5, PG_U_LOWERCASE_LETTER},
{0x00a7d6, 0x00a7d6, PG_U_UPPERCASE_LETTER},
{0x00a7d7, 0x00a7d7, PG_U_LOWERCASE_LETTER},
{0x00a7d8, 0x00a7d8, PG_U_UPPERCASE_LETTER},
{0x00a7d9, 0x00a7d9, PG_U_LOWERCASE_LETTER},
- {0x00a7da, 0x00a7f1, PG_U_UNASSIGNED},
{0x00a7f2, 0x00a7f4, PG_U_MODIFIER_LETTER},
{0x00a7f5, 0x00a7f5, PG_U_UPPERCASE_LETTER},
{0x00a7f6, 0x00a7f6, PG_U_LOWERCASE_LETTER},
@@ -2623,23 +2339,18 @@ static const pg_category_range unicode_categories[4009] =
{0x00a827, 0x00a827, PG_U_SPACING_MARK},
{0x00a828, 0x00a82b, PG_U_OTHER_SYMBOL},
{0x00a82c, 0x00a82c, PG_U_NONSPACING_MARK},
- {0x00a82d, 0x00a82f, PG_U_UNASSIGNED},
{0x00a830, 0x00a835, PG_U_OTHER_NUMBER},
{0x00a836, 0x00a837, PG_U_OTHER_SYMBOL},
{0x00a838, 0x00a838, PG_U_CURRENCY_SYMBOL},
{0x00a839, 0x00a839, PG_U_OTHER_SYMBOL},
- {0x00a83a, 0x00a83f, PG_U_UNASSIGNED},
{0x00a840, 0x00a873, PG_U_OTHER_LETTER},
{0x00a874, 0x00a877, PG_U_OTHER_PUNCTUATION},
- {0x00a878, 0x00a87f, PG_U_UNASSIGNED},
{0x00a880, 0x00a881, PG_U_SPACING_MARK},
{0x00a882, 0x00a8b3, PG_U_OTHER_LETTER},
{0x00a8b4, 0x00a8c3, PG_U_SPACING_MARK},
{0x00a8c4, 0x00a8c5, PG_U_NONSPACING_MARK},
- {0x00a8c6, 0x00a8cd, PG_U_UNASSIGNED},
{0x00a8ce, 0x00a8cf, PG_U_OTHER_PUNCTUATION},
{0x00a8d0, 0x00a8d9, PG_U_DECIMAL_NUMBER},
- {0x00a8da, 0x00a8df, PG_U_UNASSIGNED},
{0x00a8e0, 0x00a8f1, PG_U_NONSPACING_MARK},
{0x00a8f2, 0x00a8f7, PG_U_OTHER_LETTER},
{0x00a8f8, 0x00a8fa, PG_U_OTHER_PUNCTUATION},
@@ -2654,10 +2365,8 @@ static const pg_category_range unicode_categories[4009] =
{0x00a930, 0x00a946, PG_U_OTHER_LETTER},
{0x00a947, 0x00a951, PG_U_NONSPACING_MARK},
{0x00a952, 0x00a953, PG_U_SPACING_MARK},
- {0x00a954, 0x00a95e, PG_U_UNASSIGNED},
{0x00a95f, 0x00a95f, PG_U_OTHER_PUNCTUATION},
{0x00a960, 0x00a97c, PG_U_OTHER_LETTER},
- {0x00a97d, 0x00a97f, PG_U_UNASSIGNED},
{0x00a980, 0x00a982, PG_U_NONSPACING_MARK},
{0x00a983, 0x00a983, PG_U_SPACING_MARK},
{0x00a984, 0x00a9b2, PG_U_OTHER_LETTER},
@@ -2668,10 +2377,8 @@ static const pg_category_range unicode_categories[4009] =
{0x00a9bc, 0x00a9bd, PG_U_NONSPACING_MARK},
{0x00a9be, 0x00a9c0, PG_U_SPACING_MARK},
{0x00a9c1, 0x00a9cd, PG_U_OTHER_PUNCTUATION},
- {0x00a9ce, 0x00a9ce, PG_U_UNASSIGNED},
{0x00a9cf, 0x00a9cf, PG_U_MODIFIER_LETTER},
{0x00a9d0, 0x00a9d9, PG_U_DECIMAL_NUMBER},
- {0x00a9da, 0x00a9dd, PG_U_UNASSIGNED},
{0x00a9de, 0x00a9df, PG_U_OTHER_PUNCTUATION},
{0x00a9e0, 0x00a9e4, PG_U_OTHER_LETTER},
{0x00a9e5, 0x00a9e5, PG_U_NONSPACING_MARK},
@@ -2679,22 +2386,18 @@ static const pg_category_range unicode_categories[4009] =
{0x00a9e7, 0x00a9ef, PG_U_OTHER_LETTER},
{0x00a9f0, 0x00a9f9, PG_U_DECIMAL_NUMBER},
{0x00a9fa, 0x00a9fe, PG_U_OTHER_LETTER},
- {0x00a9ff, 0x00a9ff, PG_U_UNASSIGNED},
{0x00aa00, 0x00aa28, PG_U_OTHER_LETTER},
{0x00aa29, 0x00aa2e, PG_U_NONSPACING_MARK},
{0x00aa2f, 0x00aa30, PG_U_SPACING_MARK},
{0x00aa31, 0x00aa32, PG_U_NONSPACING_MARK},
{0x00aa33, 0x00aa34, PG_U_SPACING_MARK},
{0x00aa35, 0x00aa36, PG_U_NONSPACING_MARK},
- {0x00aa37, 0x00aa3f, PG_U_UNASSIGNED},
{0x00aa40, 0x00aa42, PG_U_OTHER_LETTER},
{0x00aa43, 0x00aa43, PG_U_NONSPACING_MARK},
{0x00aa44, 0x00aa4b, PG_U_OTHER_LETTER},
{0x00aa4c, 0x00aa4c, PG_U_NONSPACING_MARK},
{0x00aa4d, 0x00aa4d, PG_U_SPACING_MARK},
- {0x00aa4e, 0x00aa4f, PG_U_UNASSIGNED},
{0x00aa50, 0x00aa59, PG_U_DECIMAL_NUMBER},
- {0x00aa5a, 0x00aa5b, PG_U_UNASSIGNED},
{0x00aa5c, 0x00aa5f, PG_U_OTHER_PUNCTUATION},
{0x00aa60, 0x00aa6f, PG_U_OTHER_LETTER},
{0x00aa70, 0x00aa70, PG_U_MODIFIER_LETTER},
@@ -2715,7 +2418,6 @@ static const pg_category_range unicode_categories[4009] =
{0x00aac0, 0x00aac0, PG_U_OTHER_LETTER},
{0x00aac1, 0x00aac1, PG_U_NONSPACING_MARK},
{0x00aac2, 0x00aac2, PG_U_OTHER_LETTER},
- {0x00aac3, 0x00aada, PG_U_UNASSIGNED},
{0x00aadb, 0x00aadc, PG_U_OTHER_LETTER},
{0x00aadd, 0x00aadd, PG_U_MODIFIER_LETTER},
{0x00aade, 0x00aadf, PG_U_OTHER_PUNCTUATION},
@@ -2728,24 +2430,17 @@ static const pg_category_range unicode_categories[4009] =
{0x00aaf3, 0x00aaf4, PG_U_MODIFIER_LETTER},
{0x00aaf5, 0x00aaf5, PG_U_SPACING_MARK},
{0x00aaf6, 0x00aaf6, PG_U_NONSPACING_MARK},
- {0x00aaf7, 0x00ab00, PG_U_UNASSIGNED},
{0x00ab01, 0x00ab06, PG_U_OTHER_LETTER},
- {0x00ab07, 0x00ab08, PG_U_UNASSIGNED},
{0x00ab09, 0x00ab0e, PG_U_OTHER_LETTER},
- {0x00ab0f, 0x00ab10, PG_U_UNASSIGNED},
{0x00ab11, 0x00ab16, PG_U_OTHER_LETTER},
- {0x00ab17, 0x00ab1f, PG_U_UNASSIGNED},
{0x00ab20, 0x00ab26, PG_U_OTHER_LETTER},
- {0x00ab27, 0x00ab27, PG_U_UNASSIGNED},
{0x00ab28, 0x00ab2e, PG_U_OTHER_LETTER},
- {0x00ab2f, 0x00ab2f, PG_U_UNASSIGNED},
{0x00ab30, 0x00ab5a, PG_U_LOWERCASE_LETTER},
{0x00ab5b, 0x00ab5b, PG_U_MODIFIER_SYMBOL},
{0x00ab5c, 0x00ab5f, PG_U_MODIFIER_LETTER},
{0x00ab60, 0x00ab68, PG_U_LOWERCASE_LETTER},
{0x00ab69, 0x00ab69, PG_U_MODIFIER_LETTER},
{0x00ab6a, 0x00ab6b, PG_U_MODIFIER_SYMBOL},
- {0x00ab6c, 0x00ab6f, PG_U_UNASSIGNED},
{0x00ab70, 0x00abbf, PG_U_LOWERCASE_LETTER},
{0x00abc0, 0x00abe2, PG_U_OTHER_LETTER},
{0x00abe3, 0x00abe4, PG_U_SPACING_MARK},
@@ -2756,52 +2451,34 @@ static const pg_category_range unicode_categories[4009] =
{0x00abeb, 0x00abeb, PG_U_OTHER_PUNCTUATION},
{0x00abec, 0x00abec, PG_U_SPACING_MARK},
{0x00abed, 0x00abed, PG_U_NONSPACING_MARK},
- {0x00abee, 0x00abef, PG_U_UNASSIGNED},
{0x00abf0, 0x00abf9, PG_U_DECIMAL_NUMBER},
- {0x00abfa, 0x00abff, PG_U_UNASSIGNED},
{0x00ac00, 0x00d7a3, PG_U_OTHER_LETTER},
- {0x00d7a4, 0x00d7af, PG_U_UNASSIGNED},
{0x00d7b0, 0x00d7c6, PG_U_OTHER_LETTER},
- {0x00d7c7, 0x00d7ca, PG_U_UNASSIGNED},
{0x00d7cb, 0x00d7fb, PG_U_OTHER_LETTER},
- {0x00d7fc, 0x00d7ff, PG_U_UNASSIGNED},
{0x00d800, 0x00dfff, PG_U_SURROGATE},
{0x00e000, 0x00f8ff, PG_U_PRIVATE_USE},
{0x00f900, 0x00fa6d, PG_U_OTHER_LETTER},
- {0x00fa6e, 0x00fa6f, PG_U_UNASSIGNED},
{0x00fa70, 0x00fad9, PG_U_OTHER_LETTER},
- {0x00fada, 0x00faff, PG_U_UNASSIGNED},
{0x00fb00, 0x00fb06, PG_U_LOWERCASE_LETTER},
- {0x00fb07, 0x00fb12, PG_U_UNASSIGNED},
{0x00fb13, 0x00fb17, PG_U_LOWERCASE_LETTER},
- {0x00fb18, 0x00fb1c, PG_U_UNASSIGNED},
{0x00fb1d, 0x00fb1d, PG_U_OTHER_LETTER},
{0x00fb1e, 0x00fb1e, PG_U_NONSPACING_MARK},
{0x00fb1f, 0x00fb28, PG_U_OTHER_LETTER},
{0x00fb29, 0x00fb29, PG_U_MATH_SYMBOL},
{0x00fb2a, 0x00fb36, PG_U_OTHER_LETTER},
- {0x00fb37, 0x00fb37, PG_U_UNASSIGNED},
{0x00fb38, 0x00fb3c, PG_U_OTHER_LETTER},
- {0x00fb3d, 0x00fb3d, PG_U_UNASSIGNED},
{0x00fb3e, 0x00fb3e, PG_U_OTHER_LETTER},
- {0x00fb3f, 0x00fb3f, PG_U_UNASSIGNED},
{0x00fb40, 0x00fb41, PG_U_OTHER_LETTER},
- {0x00fb42, 0x00fb42, PG_U_UNASSIGNED},
{0x00fb43, 0x00fb44, PG_U_OTHER_LETTER},
- {0x00fb45, 0x00fb45, PG_U_UNASSIGNED},
{0x00fb46, 0x00fbb1, PG_U_OTHER_LETTER},
{0x00fbb2, 0x00fbc2, PG_U_MODIFIER_SYMBOL},
- {0x00fbc3, 0x00fbd2, PG_U_UNASSIGNED},
{0x00fbd3, 0x00fd3d, PG_U_OTHER_LETTER},
{0x00fd3e, 0x00fd3e, PG_U_CLOSE_PUNCTUATION},
{0x00fd3f, 0x00fd3f, PG_U_OPEN_PUNCTUATION},
{0x00fd40, 0x00fd4f, PG_U_OTHER_SYMBOL},
{0x00fd50, 0x00fd8f, PG_U_OTHER_LETTER},
- {0x00fd90, 0x00fd91, PG_U_UNASSIGNED},
{0x00fd92, 0x00fdc7, PG_U_OTHER_LETTER},
- {0x00fdc8, 0x00fdce, PG_U_UNASSIGNED},
{0x00fdcf, 0x00fdcf, PG_U_OTHER_SYMBOL},
- {0x00fdd0, 0x00fdef, PG_U_UNASSIGNED},
{0x00fdf0, 0x00fdfb, PG_U_OTHER_LETTER},
{0x00fdfc, 0x00fdfc, PG_U_CURRENCY_SYMBOL},
{0x00fdfd, 0x00fdff, PG_U_OTHER_SYMBOL},
@@ -2810,7 +2487,6 @@ static const pg_category_range unicode_categories[4009] =
{0x00fe17, 0x00fe17, PG_U_OPEN_PUNCTUATION},
{0x00fe18, 0x00fe18, PG_U_CLOSE_PUNCTUATION},
{0x00fe19, 0x00fe19, PG_U_OTHER_PUNCTUATION},
- {0x00fe1a, 0x00fe1f, PG_U_UNASSIGNED},
{0x00fe20, 0x00fe2f, PG_U_NONSPACING_MARK},
{0x00fe30, 0x00fe30, PG_U_OTHER_PUNCTUATION},
{0x00fe31, 0x00fe32, PG_U_DASH_PUNCTUATION},
@@ -2837,7 +2513,6 @@ static const pg_category_range unicode_categories[4009] =
{0x00fe49, 0x00fe4c, PG_U_OTHER_PUNCTUATION},
{0x00fe4d, 0x00fe4f, PG_U_CONNECTOR_PUNCTUATION},
{0x00fe50, 0x00fe52, PG_U_OTHER_PUNCTUATION},
- {0x00fe53, 0x00fe53, PG_U_UNASSIGNED},
{0x00fe54, 0x00fe57, PG_U_OTHER_PUNCTUATION},
{0x00fe58, 0x00fe58, PG_U_DASH_PUNCTUATION},
{0x00fe59, 0x00fe59, PG_U_OPEN_PUNCTUATION},
@@ -2850,17 +2525,12 @@ static const pg_category_range unicode_categories[4009] =
{0x00fe62, 0x00fe62, PG_U_MATH_SYMBOL},
{0x00fe63, 0x00fe63, PG_U_DASH_PUNCTUATION},
{0x00fe64, 0x00fe66, PG_U_MATH_SYMBOL},
- {0x00fe67, 0x00fe67, PG_U_UNASSIGNED},
{0x00fe68, 0x00fe68, PG_U_OTHER_PUNCTUATION},
{0x00fe69, 0x00fe69, PG_U_CURRENCY_SYMBOL},
{0x00fe6a, 0x00fe6b, PG_U_OTHER_PUNCTUATION},
- {0x00fe6c, 0x00fe6f, PG_U_UNASSIGNED},
{0x00fe70, 0x00fe74, PG_U_OTHER_LETTER},
- {0x00fe75, 0x00fe75, PG_U_UNASSIGNED},
{0x00fe76, 0x00fefc, PG_U_OTHER_LETTER},
- {0x00fefd, 0x00fefe, PG_U_UNASSIGNED},
{0x00feff, 0x00feff, PG_U_FORMAT},
- {0x00ff00, 0x00ff00, PG_U_UNASSIGNED},
{0x00ff01, 0x00ff03, PG_U_OTHER_PUNCTUATION},
{0x00ff04, 0x00ff04, PG_U_CURRENCY_SYMBOL},
{0x00ff05, 0x00ff07, PG_U_OTHER_PUNCTUATION},
@@ -2898,273 +2568,175 @@ static const pg_category_range unicode_categories[4009] =
{0x00ff71, 0x00ff9d, PG_U_OTHER_LETTER},
{0x00ff9e, 0x00ff9f, PG_U_MODIFIER_LETTER},
{0x00ffa0, 0x00ffbe, PG_U_OTHER_LETTER},
- {0x00ffbf, 0x00ffc1, PG_U_UNASSIGNED},
{0x00ffc2, 0x00ffc7, PG_U_OTHER_LETTER},
- {0x00ffc8, 0x00ffc9, PG_U_UNASSIGNED},
{0x00ffca, 0x00ffcf, PG_U_OTHER_LETTER},
- {0x00ffd0, 0x00ffd1, PG_U_UNASSIGNED},
{0x00ffd2, 0x00ffd7, PG_U_OTHER_LETTER},
- {0x00ffd8, 0x00ffd9, PG_U_UNASSIGNED},
{0x00ffda, 0x00ffdc, PG_U_OTHER_LETTER},
- {0x00ffdd, 0x00ffdf, PG_U_UNASSIGNED},
{0x00ffe0, 0x00ffe1, PG_U_CURRENCY_SYMBOL},
{0x00ffe2, 0x00ffe2, PG_U_MATH_SYMBOL},
{0x00ffe3, 0x00ffe3, PG_U_MODIFIER_SYMBOL},
{0x00ffe4, 0x00ffe4, PG_U_OTHER_SYMBOL},
{0x00ffe5, 0x00ffe6, PG_U_CURRENCY_SYMBOL},
- {0x00ffe7, 0x00ffe7, PG_U_UNASSIGNED},
{0x00ffe8, 0x00ffe8, PG_U_OTHER_SYMBOL},
{0x00ffe9, 0x00ffec, PG_U_MATH_SYMBOL},
{0x00ffed, 0x00ffee, PG_U_OTHER_SYMBOL},
- {0x00ffef, 0x00fff8, PG_U_UNASSIGNED},
{0x00fff9, 0x00fffb, PG_U_FORMAT},
{0x00fffc, 0x00fffd, PG_U_OTHER_SYMBOL},
- {0x00fffe, 0x00ffff, PG_U_UNASSIGNED},
{0x010000, 0x01000b, PG_U_OTHER_LETTER},
- {0x01000c, 0x01000c, PG_U_UNASSIGNED},
{0x01000d, 0x010026, PG_U_OTHER_LETTER},
- {0x010027, 0x010027, PG_U_UNASSIGNED},
{0x010028, 0x01003a, PG_U_OTHER_LETTER},
- {0x01003b, 0x01003b, PG_U_UNASSIGNED},
{0x01003c, 0x01003d, PG_U_OTHER_LETTER},
- {0x01003e, 0x01003e, PG_U_UNASSIGNED},
{0x01003f, 0x01004d, PG_U_OTHER_LETTER},
- {0x01004e, 0x01004f, PG_U_UNASSIGNED},
{0x010050, 0x01005d, PG_U_OTHER_LETTER},
- {0x01005e, 0x01007f, PG_U_UNASSIGNED},
{0x010080, 0x0100fa, PG_U_OTHER_LETTER},
- {0x0100fb, 0x0100ff, PG_U_UNASSIGNED},
{0x010100, 0x010102, PG_U_OTHER_PUNCTUATION},
- {0x010103, 0x010106, PG_U_UNASSIGNED},
{0x010107, 0x010133, PG_U_OTHER_NUMBER},
- {0x010134, 0x010136, PG_U_UNASSIGNED},
{0x010137, 0x01013f, PG_U_OTHER_SYMBOL},
{0x010140, 0x010174, PG_U_LETTER_NUMBER},
{0x010175, 0x010178, PG_U_OTHER_NUMBER},
{0x010179, 0x010189, PG_U_OTHER_SYMBOL},
{0x01018a, 0x01018b, PG_U_OTHER_NUMBER},
{0x01018c, 0x01018e, PG_U_OTHER_SYMBOL},
- {0x01018f, 0x01018f, PG_U_UNASSIGNED},
{0x010190, 0x01019c, PG_U_OTHER_SYMBOL},
- {0x01019d, 0x01019f, PG_U_UNASSIGNED},
{0x0101a0, 0x0101a0, PG_U_OTHER_SYMBOL},
- {0x0101a1, 0x0101cf, PG_U_UNASSIGNED},
{0x0101d0, 0x0101fc, PG_U_OTHER_SYMBOL},
{0x0101fd, 0x0101fd, PG_U_NONSPACING_MARK},
- {0x0101fe, 0x01027f, PG_U_UNASSIGNED},
{0x010280, 0x01029c, PG_U_OTHER_LETTER},
- {0x01029d, 0x01029f, PG_U_UNASSIGNED},
{0x0102a0, 0x0102d0, PG_U_OTHER_LETTER},
- {0x0102d1, 0x0102df, PG_U_UNASSIGNED},
{0x0102e0, 0x0102e0, PG_U_NONSPACING_MARK},
{0x0102e1, 0x0102fb, PG_U_OTHER_NUMBER},
- {0x0102fc, 0x0102ff, PG_U_UNASSIGNED},
{0x010300, 0x01031f, PG_U_OTHER_LETTER},
{0x010320, 0x010323, PG_U_OTHER_NUMBER},
- {0x010324, 0x01032c, PG_U_UNASSIGNED},
{0x01032d, 0x010340, PG_U_OTHER_LETTER},
{0x010341, 0x010341, PG_U_LETTER_NUMBER},
{0x010342, 0x010349, PG_U_OTHER_LETTER},
{0x01034a, 0x01034a, PG_U_LETTER_NUMBER},
- {0x01034b, 0x01034f, PG_U_UNASSIGNED},
{0x010350, 0x010375, PG_U_OTHER_LETTER},
{0x010376, 0x01037a, PG_U_NONSPACING_MARK},
- {0x01037b, 0x01037f, PG_U_UNASSIGNED},
{0x010380, 0x01039d, PG_U_OTHER_LETTER},
- {0x01039e, 0x01039e, PG_U_UNASSIGNED},
{0x01039f, 0x01039f, PG_U_OTHER_PUNCTUATION},
{0x0103a0, 0x0103c3, PG_U_OTHER_LETTER},
- {0x0103c4, 0x0103c7, PG_U_UNASSIGNED},
{0x0103c8, 0x0103cf, PG_U_OTHER_LETTER},
{0x0103d0, 0x0103d0, PG_U_OTHER_PUNCTUATION},
{0x0103d1, 0x0103d5, PG_U_LETTER_NUMBER},
- {0x0103d6, 0x0103ff, PG_U_UNASSIGNED},
{0x010400, 0x010427, PG_U_UPPERCASE_LETTER},
{0x010428, 0x01044f, PG_U_LOWERCASE_LETTER},
{0x010450, 0x01049d, PG_U_OTHER_LETTER},
- {0x01049e, 0x01049f, PG_U_UNASSIGNED},
{0x0104a0, 0x0104a9, PG_U_DECIMAL_NUMBER},
- {0x0104aa, 0x0104af, PG_U_UNASSIGNED},
{0x0104b0, 0x0104d3, PG_U_UPPERCASE_LETTER},
- {0x0104d4, 0x0104d7, PG_U_UNASSIGNED},
{0x0104d8, 0x0104fb, PG_U_LOWERCASE_LETTER},
- {0x0104fc, 0x0104ff, PG_U_UNASSIGNED},
{0x010500, 0x010527, PG_U_OTHER_LETTER},
- {0x010528, 0x01052f, PG_U_UNASSIGNED},
{0x010530, 0x010563, PG_U_OTHER_LETTER},
- {0x010564, 0x01056e, PG_U_UNASSIGNED},
{0x01056f, 0x01056f, PG_U_OTHER_PUNCTUATION},
{0x010570, 0x01057a, PG_U_UPPERCASE_LETTER},
- {0x01057b, 0x01057b, PG_U_UNASSIGNED},
{0x01057c, 0x01058a, PG_U_UPPERCASE_LETTER},
- {0x01058b, 0x01058b, PG_U_UNASSIGNED},
{0x01058c, 0x010592, PG_U_UPPERCASE_LETTER},
- {0x010593, 0x010593, PG_U_UNASSIGNED},
{0x010594, 0x010595, PG_U_UPPERCASE_LETTER},
- {0x010596, 0x010596, PG_U_UNASSIGNED},
{0x010597, 0x0105a1, PG_U_LOWERCASE_LETTER},
- {0x0105a2, 0x0105a2, PG_U_UNASSIGNED},
{0x0105a3, 0x0105b1, PG_U_LOWERCASE_LETTER},
- {0x0105b2, 0x0105b2, PG_U_UNASSIGNED},
{0x0105b3, 0x0105b9, PG_U_LOWERCASE_LETTER},
- {0x0105ba, 0x0105ba, PG_U_UNASSIGNED},
{0x0105bb, 0x0105bc, PG_U_LOWERCASE_LETTER},
- {0x0105bd, 0x0105ff, PG_U_UNASSIGNED},
{0x010600, 0x010736, PG_U_OTHER_LETTER},
- {0x010737, 0x01073f, PG_U_UNASSIGNED},
{0x010740, 0x010755, PG_U_OTHER_LETTER},
- {0x010756, 0x01075f, PG_U_UNASSIGNED},
{0x010760, 0x010767, PG_U_OTHER_LETTER},
- {0x010768, 0x01077f, PG_U_UNASSIGNED},
{0x010780, 0x010785, PG_U_MODIFIER_LETTER},
- {0x010786, 0x010786, PG_U_UNASSIGNED},
{0x010787, 0x0107b0, PG_U_MODIFIER_LETTER},
- {0x0107b1, 0x0107b1, PG_U_UNASSIGNED},
{0x0107b2, 0x0107ba, PG_U_MODIFIER_LETTER},
- {0x0107bb, 0x0107ff, PG_U_UNASSIGNED},
{0x010800, 0x010805, PG_U_OTHER_LETTER},
- {0x010806, 0x010807, PG_U_UNASSIGNED},
{0x010808, 0x010808, PG_U_OTHER_LETTER},
- {0x010809, 0x010809, PG_U_UNASSIGNED},
{0x01080a, 0x010835, PG_U_OTHER_LETTER},
- {0x010836, 0x010836, PG_U_UNASSIGNED},
{0x010837, 0x010838, PG_U_OTHER_LETTER},
- {0x010839, 0x01083b, PG_U_UNASSIGNED},
{0x01083c, 0x01083c, PG_U_OTHER_LETTER},
- {0x01083d, 0x01083e, PG_U_UNASSIGNED},
{0x01083f, 0x010855, PG_U_OTHER_LETTER},
- {0x010856, 0x010856, PG_U_UNASSIGNED},
{0x010857, 0x010857, PG_U_OTHER_PUNCTUATION},
{0x010858, 0x01085f, PG_U_OTHER_NUMBER},
{0x010860, 0x010876, PG_U_OTHER_LETTER},
{0x010877, 0x010878, PG_U_OTHER_SYMBOL},
{0x010879, 0x01087f, PG_U_OTHER_NUMBER},
{0x010880, 0x01089e, PG_U_OTHER_LETTER},
- {0x01089f, 0x0108a6, PG_U_UNASSIGNED},
{0x0108a7, 0x0108af, PG_U_OTHER_NUMBER},
- {0x0108b0, 0x0108df, PG_U_UNASSIGNED},
{0x0108e0, 0x0108f2, PG_U_OTHER_LETTER},
- {0x0108f3, 0x0108f3, PG_U_UNASSIGNED},
{0x0108f4, 0x0108f5, PG_U_OTHER_LETTER},
- {0x0108f6, 0x0108fa, PG_U_UNASSIGNED},
{0x0108fb, 0x0108ff, PG_U_OTHER_NUMBER},
{0x010900, 0x010915, PG_U_OTHER_LETTER},
{0x010916, 0x01091b, PG_U_OTHER_NUMBER},
- {0x01091c, 0x01091e, PG_U_UNASSIGNED},
{0x01091f, 0x01091f, PG_U_OTHER_PUNCTUATION},
{0x010920, 0x010939, PG_U_OTHER_LETTER},
- {0x01093a, 0x01093e, PG_U_UNASSIGNED},
{0x01093f, 0x01093f, PG_U_OTHER_PUNCTUATION},
- {0x010940, 0x01097f, PG_U_UNASSIGNED},
{0x010980, 0x0109b7, PG_U_OTHER_LETTER},
- {0x0109b8, 0x0109bb, PG_U_UNASSIGNED},
{0x0109bc, 0x0109bd, PG_U_OTHER_NUMBER},
{0x0109be, 0x0109bf, PG_U_OTHER_LETTER},
{0x0109c0, 0x0109cf, PG_U_OTHER_NUMBER},
- {0x0109d0, 0x0109d1, PG_U_UNASSIGNED},
{0x0109d2, 0x0109ff, PG_U_OTHER_NUMBER},
{0x010a00, 0x010a00, PG_U_OTHER_LETTER},
{0x010a01, 0x010a03, PG_U_NONSPACING_MARK},
- {0x010a04, 0x010a04, PG_U_UNASSIGNED},
{0x010a05, 0x010a06, PG_U_NONSPACING_MARK},
- {0x010a07, 0x010a0b, PG_U_UNASSIGNED},
{0x010a0c, 0x010a0f, PG_U_NONSPACING_MARK},
{0x010a10, 0x010a13, PG_U_OTHER_LETTER},
- {0x010a14, 0x010a14, PG_U_UNASSIGNED},
{0x010a15, 0x010a17, PG_U_OTHER_LETTER},
- {0x010a18, 0x010a18, PG_U_UNASSIGNED},
{0x010a19, 0x010a35, PG_U_OTHER_LETTER},
- {0x010a36, 0x010a37, PG_U_UNASSIGNED},
{0x010a38, 0x010a3a, PG_U_NONSPACING_MARK},
- {0x010a3b, 0x010a3e, PG_U_UNASSIGNED},
{0x010a3f, 0x010a3f, PG_U_NONSPACING_MARK},
{0x010a40, 0x010a48, PG_U_OTHER_NUMBER},
- {0x010a49, 0x010a4f, PG_U_UNASSIGNED},
{0x010a50, 0x010a58, PG_U_OTHER_PUNCTUATION},
- {0x010a59, 0x010a5f, PG_U_UNASSIGNED},
{0x010a60, 0x010a7c, PG_U_OTHER_LETTER},
{0x010a7d, 0x010a7e, PG_U_OTHER_NUMBER},
{0x010a7f, 0x010a7f, PG_U_OTHER_PUNCTUATION},
{0x010a80, 0x010a9c, PG_U_OTHER_LETTER},
{0x010a9d, 0x010a9f, PG_U_OTHER_NUMBER},
- {0x010aa0, 0x010abf, PG_U_UNASSIGNED},
{0x010ac0, 0x010ac7, PG_U_OTHER_LETTER},
{0x010ac8, 0x010ac8, PG_U_OTHER_SYMBOL},
{0x010ac9, 0x010ae4, PG_U_OTHER_LETTER},
{0x010ae5, 0x010ae6, PG_U_NONSPACING_MARK},
- {0x010ae7, 0x010aea, PG_U_UNASSIGNED},
{0x010aeb, 0x010aef, PG_U_OTHER_NUMBER},
{0x010af0, 0x010af6, PG_U_OTHER_PUNCTUATION},
- {0x010af7, 0x010aff, PG_U_UNASSIGNED},
{0x010b00, 0x010b35, PG_U_OTHER_LETTER},
- {0x010b36, 0x010b38, PG_U_UNASSIGNED},
{0x010b39, 0x010b3f, PG_U_OTHER_PUNCTUATION},
{0x010b40, 0x010b55, PG_U_OTHER_LETTER},
- {0x010b56, 0x010b57, PG_U_UNASSIGNED},
{0x010b58, 0x010b5f, PG_U_OTHER_NUMBER},
{0x010b60, 0x010b72, PG_U_OTHER_LETTER},
- {0x010b73, 0x010b77, PG_U_UNASSIGNED},
{0x010b78, 0x010b7f, PG_U_OTHER_NUMBER},
{0x010b80, 0x010b91, PG_U_OTHER_LETTER},
- {0x010b92, 0x010b98, PG_U_UNASSIGNED},
{0x010b99, 0x010b9c, PG_U_OTHER_PUNCTUATION},
- {0x010b9d, 0x010ba8, PG_U_UNASSIGNED},
{0x010ba9, 0x010baf, PG_U_OTHER_NUMBER},
- {0x010bb0, 0x010bff, PG_U_UNASSIGNED},
{0x010c00, 0x010c48, PG_U_OTHER_LETTER},
- {0x010c49, 0x010c7f, PG_U_UNASSIGNED},
{0x010c80, 0x010cb2, PG_U_UPPERCASE_LETTER},
- {0x010cb3, 0x010cbf, PG_U_UNASSIGNED},
{0x010cc0, 0x010cf2, PG_U_LOWERCASE_LETTER},
- {0x010cf3, 0x010cf9, PG_U_UNASSIGNED},
{0x010cfa, 0x010cff, PG_U_OTHER_NUMBER},
{0x010d00, 0x010d23, PG_U_OTHER_LETTER},
{0x010d24, 0x010d27, PG_U_NONSPACING_MARK},
- {0x010d28, 0x010d2f, PG_U_UNASSIGNED},
{0x010d30, 0x010d39, PG_U_DECIMAL_NUMBER},
- {0x010d3a, 0x010e5f, PG_U_UNASSIGNED},
{0x010e60, 0x010e7e, PG_U_OTHER_NUMBER},
- {0x010e7f, 0x010e7f, PG_U_UNASSIGNED},
{0x010e80, 0x010ea9, PG_U_OTHER_LETTER},
- {0x010eaa, 0x010eaa, PG_U_UNASSIGNED},
{0x010eab, 0x010eac, PG_U_NONSPACING_MARK},
{0x010ead, 0x010ead, PG_U_DASH_PUNCTUATION},
- {0x010eae, 0x010eaf, PG_U_UNASSIGNED},
{0x010eb0, 0x010eb1, PG_U_OTHER_LETTER},
- {0x010eb2, 0x010efc, PG_U_UNASSIGNED},
{0x010efd, 0x010eff, PG_U_NONSPACING_MARK},
{0x010f00, 0x010f1c, PG_U_OTHER_LETTER},
{0x010f1d, 0x010f26, PG_U_OTHER_NUMBER},
{0x010f27, 0x010f27, PG_U_OTHER_LETTER},
- {0x010f28, 0x010f2f, PG_U_UNASSIGNED},
{0x010f30, 0x010f45, PG_U_OTHER_LETTER},
{0x010f46, 0x010f50, PG_U_NONSPACING_MARK},
{0x010f51, 0x010f54, PG_U_OTHER_NUMBER},
{0x010f55, 0x010f59, PG_U_OTHER_PUNCTUATION},
- {0x010f5a, 0x010f6f, PG_U_UNASSIGNED},
{0x010f70, 0x010f81, PG_U_OTHER_LETTER},
{0x010f82, 0x010f85, PG_U_NONSPACING_MARK},
{0x010f86, 0x010f89, PG_U_OTHER_PUNCTUATION},
- {0x010f8a, 0x010faf, PG_U_UNASSIGNED},
{0x010fb0, 0x010fc4, PG_U_OTHER_LETTER},
{0x010fc5, 0x010fcb, PG_U_OTHER_NUMBER},
- {0x010fcc, 0x010fdf, PG_U_UNASSIGNED},
{0x010fe0, 0x010ff6, PG_U_OTHER_LETTER},
- {0x010ff7, 0x010fff, PG_U_UNASSIGNED},
{0x011000, 0x011000, PG_U_SPACING_MARK},
{0x011001, 0x011001, PG_U_NONSPACING_MARK},
{0x011002, 0x011002, PG_U_SPACING_MARK},
{0x011003, 0x011037, PG_U_OTHER_LETTER},
{0x011038, 0x011046, PG_U_NONSPACING_MARK},
{0x011047, 0x01104d, PG_U_OTHER_PUNCTUATION},
- {0x01104e, 0x011051, PG_U_UNASSIGNED},
{0x011052, 0x011065, PG_U_OTHER_NUMBER},
{0x011066, 0x01106f, PG_U_DECIMAL_NUMBER},
{0x011070, 0x011070, PG_U_NONSPACING_MARK},
{0x011071, 0x011072, PG_U_OTHER_LETTER},
{0x011073, 0x011074, PG_U_NONSPACING_MARK},
{0x011075, 0x011075, PG_U_OTHER_LETTER},
- {0x011076, 0x01107e, PG_U_UNASSIGNED},
{0x01107f, 0x011081, PG_U_NONSPACING_MARK},
{0x011082, 0x011082, PG_U_SPACING_MARK},
{0x011083, 0x0110af, PG_U_OTHER_LETTER},
@@ -3176,30 +2748,23 @@ static const pg_category_range unicode_categories[4009] =
{0x0110bd, 0x0110bd, PG_U_FORMAT},
{0x0110be, 0x0110c1, PG_U_OTHER_PUNCTUATION},
{0x0110c2, 0x0110c2, PG_U_NONSPACING_MARK},
- {0x0110c3, 0x0110cc, PG_U_UNASSIGNED},
{0x0110cd, 0x0110cd, PG_U_FORMAT},
- {0x0110ce, 0x0110cf, PG_U_UNASSIGNED},
{0x0110d0, 0x0110e8, PG_U_OTHER_LETTER},
- {0x0110e9, 0x0110ef, PG_U_UNASSIGNED},
{0x0110f0, 0x0110f9, PG_U_DECIMAL_NUMBER},
- {0x0110fa, 0x0110ff, PG_U_UNASSIGNED},
{0x011100, 0x011102, PG_U_NONSPACING_MARK},
{0x011103, 0x011126, PG_U_OTHER_LETTER},
{0x011127, 0x01112b, PG_U_NONSPACING_MARK},
{0x01112c, 0x01112c, PG_U_SPACING_MARK},
{0x01112d, 0x011134, PG_U_NONSPACING_MARK},
- {0x011135, 0x011135, PG_U_UNASSIGNED},
{0x011136, 0x01113f, PG_U_DECIMAL_NUMBER},
{0x011140, 0x011143, PG_U_OTHER_PUNCTUATION},
{0x011144, 0x011144, PG_U_OTHER_LETTER},
{0x011145, 0x011146, PG_U_SPACING_MARK},
{0x011147, 0x011147, PG_U_OTHER_LETTER},
- {0x011148, 0x01114f, PG_U_UNASSIGNED},
{0x011150, 0x011172, PG_U_OTHER_LETTER},
{0x011173, 0x011173, PG_U_NONSPACING_MARK},
{0x011174, 0x011175, PG_U_OTHER_PUNCTUATION},
{0x011176, 0x011176, PG_U_OTHER_LETTER},
- {0x011177, 0x01117f, PG_U_UNASSIGNED},
{0x011180, 0x011181, PG_U_NONSPACING_MARK},
{0x011182, 0x011182, PG_U_SPACING_MARK},
{0x011183, 0x0111b2, PG_U_OTHER_LETTER},
@@ -3217,11 +2782,8 @@ static const pg_category_range unicode_categories[4009] =
{0x0111db, 0x0111db, PG_U_OTHER_PUNCTUATION},
{0x0111dc, 0x0111dc, PG_U_OTHER_LETTER},
{0x0111dd, 0x0111df, PG_U_OTHER_PUNCTUATION},
- {0x0111e0, 0x0111e0, PG_U_UNASSIGNED},
{0x0111e1, 0x0111f4, PG_U_OTHER_NUMBER},
- {0x0111f5, 0x0111ff, PG_U_UNASSIGNED},
{0x011200, 0x011211, PG_U_OTHER_LETTER},
- {0x011212, 0x011212, PG_U_UNASSIGNED},
{0x011213, 0x01122b, PG_U_OTHER_LETTER},
{0x01122c, 0x01122e, PG_U_SPACING_MARK},
{0x01122f, 0x011231, PG_U_NONSPACING_MARK},
@@ -3233,61 +2795,38 @@ static const pg_category_range unicode_categories[4009] =
{0x01123e, 0x01123e, PG_U_NONSPACING_MARK},
{0x01123f, 0x011240, PG_U_OTHER_LETTER},
{0x011241, 0x011241, PG_U_NONSPACING_MARK},
- {0x011242, 0x01127f, PG_U_UNASSIGNED},
{0x011280, 0x011286, PG_U_OTHER_LETTER},
- {0x011287, 0x011287, PG_U_UNASSIGNED},
{0x011288, 0x011288, PG_U_OTHER_LETTER},
- {0x011289, 0x011289, PG_U_UNASSIGNED},
{0x01128a, 0x01128d, PG_U_OTHER_LETTER},
- {0x01128e, 0x01128e, PG_U_UNASSIGNED},
{0x01128f, 0x01129d, PG_U_OTHER_LETTER},
- {0x01129e, 0x01129e, PG_U_UNASSIGNED},
{0x01129f, 0x0112a8, PG_U_OTHER_LETTER},
{0x0112a9, 0x0112a9, PG_U_OTHER_PUNCTUATION},
- {0x0112aa, 0x0112af, PG_U_UNASSIGNED},
{0x0112b0, 0x0112de, PG_U_OTHER_LETTER},
{0x0112df, 0x0112df, PG_U_NONSPACING_MARK},
{0x0112e0, 0x0112e2, PG_U_SPACING_MARK},
{0x0112e3, 0x0112ea, PG_U_NONSPACING_MARK},
- {0x0112eb, 0x0112ef, PG_U_UNASSIGNED},
{0x0112f0, 0x0112f9, PG_U_DECIMAL_NUMBER},
- {0x0112fa, 0x0112ff, PG_U_UNASSIGNED},
{0x011300, 0x011301, PG_U_NONSPACING_MARK},
{0x011302, 0x011303, PG_U_SPACING_MARK},
- {0x011304, 0x011304, PG_U_UNASSIGNED},
{0x011305, 0x01130c, PG_U_OTHER_LETTER},
- {0x01130d, 0x01130e, PG_U_UNASSIGNED},
{0x01130f, 0x011310, PG_U_OTHER_LETTER},
- {0x011311, 0x011312, PG_U_UNASSIGNED},
{0x011313, 0x011328, PG_U_OTHER_LETTER},
- {0x011329, 0x011329, PG_U_UNASSIGNED},
{0x01132a, 0x011330, PG_U_OTHER_LETTER},
- {0x011331, 0x011331, PG_U_UNASSIGNED},
{0x011332, 0x011333, PG_U_OTHER_LETTER},
- {0x011334, 0x011334, PG_U_UNASSIGNED},
{0x011335, 0x011339, PG_U_OTHER_LETTER},
- {0x01133a, 0x01133a, PG_U_UNASSIGNED},
{0x01133b, 0x01133c, PG_U_NONSPACING_MARK},
{0x01133d, 0x01133d, PG_U_OTHER_LETTER},
{0x01133e, 0x01133f, PG_U_SPACING_MARK},
{0x011340, 0x011340, PG_U_NONSPACING_MARK},
{0x011341, 0x011344, PG_U_SPACING_MARK},
- {0x011345, 0x011346, PG_U_UNASSIGNED},
{0x011347, 0x011348, PG_U_SPACING_MARK},
- {0x011349, 0x01134a, PG_U_UNASSIGNED},
{0x01134b, 0x01134d, PG_U_SPACING_MARK},
- {0x01134e, 0x01134f, PG_U_UNASSIGNED},
{0x011350, 0x011350, PG_U_OTHER_LETTER},
- {0x011351, 0x011356, PG_U_UNASSIGNED},
{0x011357, 0x011357, PG_U_SPACING_MARK},
- {0x011358, 0x01135c, PG_U_UNASSIGNED},
{0x01135d, 0x011361, PG_U_OTHER_LETTER},
{0x011362, 0x011363, PG_U_SPACING_MARK},
- {0x011364, 0x011365, PG_U_UNASSIGNED},
{0x011366, 0x01136c, PG_U_NONSPACING_MARK},
- {0x01136d, 0x01136f, PG_U_UNASSIGNED},
{0x011370, 0x011374, PG_U_NONSPACING_MARK},
- {0x011375, 0x0113ff, PG_U_UNASSIGNED},
{0x011400, 0x011434, PG_U_OTHER_LETTER},
{0x011435, 0x011437, PG_U_SPACING_MARK},
{0x011438, 0x01143f, PG_U_NONSPACING_MARK},
@@ -3299,11 +2838,9 @@ static const pg_category_range unicode_categories[4009] =
{0x01144b, 0x01144f, PG_U_OTHER_PUNCTUATION},
{0x011450, 0x011459, PG_U_DECIMAL_NUMBER},
{0x01145a, 0x01145b, PG_U_OTHER_PUNCTUATION},
- {0x01145c, 0x01145c, PG_U_UNASSIGNED},
{0x01145d, 0x01145d, PG_U_OTHER_PUNCTUATION},
{0x01145e, 0x01145e, PG_U_NONSPACING_MARK},
{0x01145f, 0x011461, PG_U_OTHER_LETTER},
- {0x011462, 0x01147f, PG_U_UNASSIGNED},
{0x011480, 0x0114af, PG_U_OTHER_LETTER},
{0x0114b0, 0x0114b2, PG_U_SPACING_MARK},
{0x0114b3, 0x0114b8, PG_U_NONSPACING_MARK},
@@ -3316,13 +2853,10 @@ static const pg_category_range unicode_categories[4009] =
{0x0114c4, 0x0114c5, PG_U_OTHER_LETTER},
{0x0114c6, 0x0114c6, PG_U_OTHER_PUNCTUATION},
{0x0114c7, 0x0114c7, PG_U_OTHER_LETTER},
- {0x0114c8, 0x0114cf, PG_U_UNASSIGNED},
{0x0114d0, 0x0114d9, PG_U_DECIMAL_NUMBER},
- {0x0114da, 0x01157f, PG_U_UNASSIGNED},
{0x011580, 0x0115ae, PG_U_OTHER_LETTER},
{0x0115af, 0x0115b1, PG_U_SPACING_MARK},
{0x0115b2, 0x0115b5, PG_U_NONSPACING_MARK},
- {0x0115b6, 0x0115b7, PG_U_UNASSIGNED},
{0x0115b8, 0x0115bb, PG_U_SPACING_MARK},
{0x0115bc, 0x0115bd, PG_U_NONSPACING_MARK},
{0x0115be, 0x0115be, PG_U_SPACING_MARK},
@@ -3330,7 +2864,6 @@ static const pg_category_range unicode_categories[4009] =
{0x0115c1, 0x0115d7, PG_U_OTHER_PUNCTUATION},
{0x0115d8, 0x0115db, PG_U_OTHER_LETTER},
{0x0115dc, 0x0115dd, PG_U_NONSPACING_MARK},
- {0x0115de, 0x0115ff, PG_U_UNASSIGNED},
{0x011600, 0x01162f, PG_U_OTHER_LETTER},
{0x011630, 0x011632, PG_U_SPACING_MARK},
{0x011633, 0x01163a, PG_U_NONSPACING_MARK},
@@ -3340,11 +2873,8 @@ static const pg_category_range unicode_categories[4009] =
{0x01163f, 0x011640, PG_U_NONSPACING_MARK},
{0x011641, 0x011643, PG_U_OTHER_PUNCTUATION},
{0x011644, 0x011644, PG_U_OTHER_LETTER},
- {0x011645, 0x01164f, PG_U_UNASSIGNED},
{0x011650, 0x011659, PG_U_DECIMAL_NUMBER},
- {0x01165a, 0x01165f, PG_U_UNASSIGNED},
{0x011660, 0x01166c, PG_U_OTHER_PUNCTUATION},
- {0x01166d, 0x01167f, PG_U_UNASSIGNED},
{0x011680, 0x0116aa, PG_U_OTHER_LETTER},
{0x0116ab, 0x0116ab, PG_U_NONSPACING_MARK},
{0x0116ac, 0x0116ac, PG_U_SPACING_MARK},
@@ -3355,48 +2885,35 @@ static const pg_category_range unicode_categories[4009] =
{0x0116b7, 0x0116b7, PG_U_NONSPACING_MARK},
{0x0116b8, 0x0116b8, PG_U_OTHER_LETTER},
{0x0116b9, 0x0116b9, PG_U_OTHER_PUNCTUATION},
- {0x0116ba, 0x0116bf, PG_U_UNASSIGNED},
{0x0116c0, 0x0116c9, PG_U_DECIMAL_NUMBER},
- {0x0116ca, 0x0116ff, PG_U_UNASSIGNED},
{0x011700, 0x01171a, PG_U_OTHER_LETTER},
- {0x01171b, 0x01171c, PG_U_UNASSIGNED},
{0x01171d, 0x01171f, PG_U_NONSPACING_MARK},
{0x011720, 0x011721, PG_U_SPACING_MARK},
{0x011722, 0x011725, PG_U_NONSPACING_MARK},
{0x011726, 0x011726, PG_U_SPACING_MARK},
{0x011727, 0x01172b, PG_U_NONSPACING_MARK},
- {0x01172c, 0x01172f, PG_U_UNASSIGNED},
{0x011730, 0x011739, PG_U_DECIMAL_NUMBER},
{0x01173a, 0x01173b, PG_U_OTHER_NUMBER},
{0x01173c, 0x01173e, PG_U_OTHER_PUNCTUATION},
{0x01173f, 0x01173f, PG_U_OTHER_SYMBOL},
{0x011740, 0x011746, PG_U_OTHER_LETTER},
- {0x011747, 0x0117ff, PG_U_UNASSIGNED},
{0x011800, 0x01182b, PG_U_OTHER_LETTER},
{0x01182c, 0x01182e, PG_U_SPACING_MARK},
{0x01182f, 0x011837, PG_U_NONSPACING_MARK},
{0x011838, 0x011838, PG_U_SPACING_MARK},
{0x011839, 0x01183a, PG_U_NONSPACING_MARK},
{0x01183b, 0x01183b, PG_U_OTHER_PUNCTUATION},
- {0x01183c, 0x01189f, PG_U_UNASSIGNED},
{0x0118a0, 0x0118bf, PG_U_UPPERCASE_LETTER},
{0x0118c0, 0x0118df, PG_U_LOWERCASE_LETTER},
{0x0118e0, 0x0118e9, PG_U_DECIMAL_NUMBER},
{0x0118ea, 0x0118f2, PG_U_OTHER_NUMBER},
- {0x0118f3, 0x0118fe, PG_U_UNASSIGNED},
{0x0118ff, 0x011906, PG_U_OTHER_LETTER},
- {0x011907, 0x011908, PG_U_UNASSIGNED},
{0x011909, 0x011909, PG_U_OTHER_LETTER},
- {0x01190a, 0x01190b, PG_U_UNASSIGNED},
{0x01190c, 0x011913, PG_U_OTHER_LETTER},
- {0x011914, 0x011914, PG_U_UNASSIGNED},
{0x011915, 0x011916, PG_U_OTHER_LETTER},
- {0x011917, 0x011917, PG_U_UNASSIGNED},
{0x011918, 0x01192f, PG_U_OTHER_LETTER},
{0x011930, 0x011935, PG_U_SPACING_MARK},
- {0x011936, 0x011936, PG_U_UNASSIGNED},
{0x011937, 0x011938, PG_U_SPACING_MARK},
- {0x011939, 0x01193a, PG_U_UNASSIGNED},
{0x01193b, 0x01193c, PG_U_NONSPACING_MARK},
{0x01193d, 0x01193d, PG_U_SPACING_MARK},
{0x01193e, 0x01193e, PG_U_NONSPACING_MARK},
@@ -3406,15 +2923,11 @@ static const pg_category_range unicode_categories[4009] =
{0x011942, 0x011942, PG_U_SPACING_MARK},
{0x011943, 0x011943, PG_U_NONSPACING_MARK},
{0x011944, 0x011946, PG_U_OTHER_PUNCTUATION},
- {0x011947, 0x01194f, PG_U_UNASSIGNED},
{0x011950, 0x011959, PG_U_DECIMAL_NUMBER},
- {0x01195a, 0x01199f, PG_U_UNASSIGNED},
{0x0119a0, 0x0119a7, PG_U_OTHER_LETTER},
- {0x0119a8, 0x0119a9, PG_U_UNASSIGNED},
{0x0119aa, 0x0119d0, PG_U_OTHER_LETTER},
{0x0119d1, 0x0119d3, PG_U_SPACING_MARK},
{0x0119d4, 0x0119d7, PG_U_NONSPACING_MARK},
- {0x0119d8, 0x0119d9, PG_U_UNASSIGNED},
{0x0119da, 0x0119db, PG_U_NONSPACING_MARK},
{0x0119dc, 0x0119df, PG_U_SPACING_MARK},
{0x0119e0, 0x0119e0, PG_U_NONSPACING_MARK},
@@ -3422,7 +2935,6 @@ static const pg_category_range unicode_categories[4009] =
{0x0119e2, 0x0119e2, PG_U_OTHER_PUNCTUATION},
{0x0119e3, 0x0119e3, PG_U_OTHER_LETTER},
{0x0119e4, 0x0119e4, PG_U_SPACING_MARK},
- {0x0119e5, 0x0119ff, PG_U_UNASSIGNED},
{0x011a00, 0x011a00, PG_U_OTHER_LETTER},
{0x011a01, 0x011a0a, PG_U_NONSPACING_MARK},
{0x011a0b, 0x011a32, PG_U_OTHER_LETTER},
@@ -3432,7 +2944,6 @@ static const pg_category_range unicode_categories[4009] =
{0x011a3b, 0x011a3e, PG_U_NONSPACING_MARK},
{0x011a3f, 0x011a46, PG_U_OTHER_PUNCTUATION},
{0x011a47, 0x011a47, PG_U_NONSPACING_MARK},
- {0x011a48, 0x011a4f, PG_U_UNASSIGNED},
{0x011a50, 0x011a50, PG_U_OTHER_LETTER},
{0x011a51, 0x011a56, PG_U_NONSPACING_MARK},
{0x011a57, 0x011a58, PG_U_SPACING_MARK},
@@ -3444,136 +2955,93 @@ static const pg_category_range unicode_categories[4009] =
{0x011a9a, 0x011a9c, PG_U_OTHER_PUNCTUATION},
{0x011a9d, 0x011a9d, PG_U_OTHER_LETTER},
{0x011a9e, 0x011aa2, PG_U_OTHER_PUNCTUATION},
- {0x011aa3, 0x011aaf, PG_U_UNASSIGNED},
{0x011ab0, 0x011af8, PG_U_OTHER_LETTER},
- {0x011af9, 0x011aff, PG_U_UNASSIGNED},
{0x011b00, 0x011b09, PG_U_OTHER_PUNCTUATION},
- {0x011b0a, 0x011bff, PG_U_UNASSIGNED},
{0x011c00, 0x011c08, PG_U_OTHER_LETTER},
- {0x011c09, 0x011c09, PG_U_UNASSIGNED},
{0x011c0a, 0x011c2e, PG_U_OTHER_LETTER},
{0x011c2f, 0x011c2f, PG_U_SPACING_MARK},
{0x011c30, 0x011c36, PG_U_NONSPACING_MARK},
- {0x011c37, 0x011c37, PG_U_UNASSIGNED},
{0x011c38, 0x011c3d, PG_U_NONSPACING_MARK},
{0x011c3e, 0x011c3e, PG_U_SPACING_MARK},
{0x011c3f, 0x011c3f, PG_U_NONSPACING_MARK},
{0x011c40, 0x011c40, PG_U_OTHER_LETTER},
{0x011c41, 0x011c45, PG_U_OTHER_PUNCTUATION},
- {0x011c46, 0x011c4f, PG_U_UNASSIGNED},
{0x011c50, 0x011c59, PG_U_DECIMAL_NUMBER},
{0x011c5a, 0x011c6c, PG_U_OTHER_NUMBER},
- {0x011c6d, 0x011c6f, PG_U_UNASSIGNED},
{0x011c70, 0x011c71, PG_U_OTHER_PUNCTUATION},
{0x011c72, 0x011c8f, PG_U_OTHER_LETTER},
- {0x011c90, 0x011c91, PG_U_UNASSIGNED},
{0x011c92, 0x011ca7, PG_U_NONSPACING_MARK},
- {0x011ca8, 0x011ca8, PG_U_UNASSIGNED},
{0x011ca9, 0x011ca9, PG_U_SPACING_MARK},
{0x011caa, 0x011cb0, PG_U_NONSPACING_MARK},
{0x011cb1, 0x011cb1, PG_U_SPACING_MARK},
{0x011cb2, 0x011cb3, PG_U_NONSPACING_MARK},
{0x011cb4, 0x011cb4, PG_U_SPACING_MARK},
{0x011cb5, 0x011cb6, PG_U_NONSPACING_MARK},
- {0x011cb7, 0x011cff, PG_U_UNASSIGNED},
{0x011d00, 0x011d06, PG_U_OTHER_LETTER},
- {0x011d07, 0x011d07, PG_U_UNASSIGNED},
{0x011d08, 0x011d09, PG_U_OTHER_LETTER},
- {0x011d0a, 0x011d0a, PG_U_UNASSIGNED},
{0x011d0b, 0x011d30, PG_U_OTHER_LETTER},
{0x011d31, 0x011d36, PG_U_NONSPACING_MARK},
- {0x011d37, 0x011d39, PG_U_UNASSIGNED},
{0x011d3a, 0x011d3a, PG_U_NONSPACING_MARK},
- {0x011d3b, 0x011d3b, PG_U_UNASSIGNED},
{0x011d3c, 0x011d3d, PG_U_NONSPACING_MARK},
- {0x011d3e, 0x011d3e, PG_U_UNASSIGNED},
{0x011d3f, 0x011d45, PG_U_NONSPACING_MARK},
{0x011d46, 0x011d46, PG_U_OTHER_LETTER},
{0x011d47, 0x011d47, PG_U_NONSPACING_MARK},
- {0x011d48, 0x011d4f, PG_U_UNASSIGNED},
{0x011d50, 0x011d59, PG_U_DECIMAL_NUMBER},
- {0x011d5a, 0x011d5f, PG_U_UNASSIGNED},
{0x011d60, 0x011d65, PG_U_OTHER_LETTER},
- {0x011d66, 0x011d66, PG_U_UNASSIGNED},
{0x011d67, 0x011d68, PG_U_OTHER_LETTER},
- {0x011d69, 0x011d69, PG_U_UNASSIGNED},
{0x011d6a, 0x011d89, PG_U_OTHER_LETTER},
{0x011d8a, 0x011d8e, PG_U_SPACING_MARK},
- {0x011d8f, 0x011d8f, PG_U_UNASSIGNED},
{0x011d90, 0x011d91, PG_U_NONSPACING_MARK},
- {0x011d92, 0x011d92, PG_U_UNASSIGNED},
{0x011d93, 0x011d94, PG_U_SPACING_MARK},
{0x011d95, 0x011d95, PG_U_NONSPACING_MARK},
{0x011d96, 0x011d96, PG_U_SPACING_MARK},
{0x011d97, 0x011d97, PG_U_NONSPACING_MARK},
{0x011d98, 0x011d98, PG_U_OTHER_LETTER},
- {0x011d99, 0x011d9f, PG_U_UNASSIGNED},
{0x011da0, 0x011da9, PG_U_DECIMAL_NUMBER},
- {0x011daa, 0x011edf, PG_U_UNASSIGNED},
{0x011ee0, 0x011ef2, PG_U_OTHER_LETTER},
{0x011ef3, 0x011ef4, PG_U_NONSPACING_MARK},
{0x011ef5, 0x011ef6, PG_U_SPACING_MARK},
{0x011ef7, 0x011ef8, PG_U_OTHER_PUNCTUATION},
- {0x011ef9, 0x011eff, PG_U_UNASSIGNED},
{0x011f00, 0x011f01, PG_U_NONSPACING_MARK},
{0x011f02, 0x011f02, PG_U_OTHER_LETTER},
{0x011f03, 0x011f03, PG_U_SPACING_MARK},
{0x011f04, 0x011f10, PG_U_OTHER_LETTER},
- {0x011f11, 0x011f11, PG_U_UNASSIGNED},
{0x011f12, 0x011f33, PG_U_OTHER_LETTER},
{0x011f34, 0x011f35, PG_U_SPACING_MARK},
{0x011f36, 0x011f3a, PG_U_NONSPACING_MARK},
- {0x011f3b, 0x011f3d, PG_U_UNASSIGNED},
{0x011f3e, 0x011f3f, PG_U_SPACING_MARK},
{0x011f40, 0x011f40, PG_U_NONSPACING_MARK},
{0x011f41, 0x011f41, PG_U_SPACING_MARK},
{0x011f42, 0x011f42, PG_U_NONSPACING_MARK},
{0x011f43, 0x011f4f, PG_U_OTHER_PUNCTUATION},
{0x011f50, 0x011f59, PG_U_DECIMAL_NUMBER},
- {0x011f5a, 0x011faf, PG_U_UNASSIGNED},
{0x011fb0, 0x011fb0, PG_U_OTHER_LETTER},
- {0x011fb1, 0x011fbf, PG_U_UNASSIGNED},
{0x011fc0, 0x011fd4, PG_U_OTHER_NUMBER},
{0x011fd5, 0x011fdc, PG_U_OTHER_SYMBOL},
{0x011fdd, 0x011fe0, PG_U_CURRENCY_SYMBOL},
{0x011fe1, 0x011ff1, PG_U_OTHER_SYMBOL},
- {0x011ff2, 0x011ffe, PG_U_UNASSIGNED},
{0x011fff, 0x011fff, PG_U_OTHER_PUNCTUATION},
{0x012000, 0x012399, PG_U_OTHER_LETTER},
- {0x01239a, 0x0123ff, PG_U_UNASSIGNED},
{0x012400, 0x01246e, PG_U_LETTER_NUMBER},
- {0x01246f, 0x01246f, PG_U_UNASSIGNED},
{0x012470, 0x012474, PG_U_OTHER_PUNCTUATION},
- {0x012475, 0x01247f, PG_U_UNASSIGNED},
{0x012480, 0x012543, PG_U_OTHER_LETTER},
- {0x012544, 0x012f8f, PG_U_UNASSIGNED},
{0x012f90, 0x012ff0, PG_U_OTHER_LETTER},
{0x012ff1, 0x012ff2, PG_U_OTHER_PUNCTUATION},
- {0x012ff3, 0x012fff, PG_U_UNASSIGNED},
{0x013000, 0x01342f, PG_U_OTHER_LETTER},
{0x013430, 0x01343f, PG_U_FORMAT},
{0x013440, 0x013440, PG_U_NONSPACING_MARK},
{0x013441, 0x013446, PG_U_OTHER_LETTER},
{0x013447, 0x013455, PG_U_NONSPACING_MARK},
- {0x013456, 0x0143ff, PG_U_UNASSIGNED},
{0x014400, 0x014646, PG_U_OTHER_LETTER},
- {0x014647, 0x0167ff, PG_U_UNASSIGNED},
{0x016800, 0x016a38, PG_U_OTHER_LETTER},
- {0x016a39, 0x016a3f, PG_U_UNASSIGNED},
{0x016a40, 0x016a5e, PG_U_OTHER_LETTER},
- {0x016a5f, 0x016a5f, PG_U_UNASSIGNED},
{0x016a60, 0x016a69, PG_U_DECIMAL_NUMBER},
- {0x016a6a, 0x016a6d, PG_U_UNASSIGNED},
{0x016a6e, 0x016a6f, PG_U_OTHER_PUNCTUATION},
{0x016a70, 0x016abe, PG_U_OTHER_LETTER},
- {0x016abf, 0x016abf, PG_U_UNASSIGNED},
{0x016ac0, 0x016ac9, PG_U_DECIMAL_NUMBER},
- {0x016aca, 0x016acf, PG_U_UNASSIGNED},
{0x016ad0, 0x016aed, PG_U_OTHER_LETTER},
- {0x016aee, 0x016aef, PG_U_UNASSIGNED},
{0x016af0, 0x016af4, PG_U_NONSPACING_MARK},
{0x016af5, 0x016af5, PG_U_OTHER_PUNCTUATION},
- {0x016af6, 0x016aff, PG_U_UNASSIGNED},
{0x016b00, 0x016b2f, PG_U_OTHER_LETTER},
{0x016b30, 0x016b36, PG_U_NONSPACING_MARK},
{0x016b37, 0x016b3b, PG_U_OTHER_PUNCTUATION},
@@ -3581,83 +3049,50 @@ static const pg_category_range unicode_categories[4009] =
{0x016b40, 0x016b43, PG_U_MODIFIER_LETTER},
{0x016b44, 0x016b44, PG_U_OTHER_PUNCTUATION},
{0x016b45, 0x016b45, PG_U_OTHER_SYMBOL},
- {0x016b46, 0x016b4f, PG_U_UNASSIGNED},
{0x016b50, 0x016b59, PG_U_DECIMAL_NUMBER},
- {0x016b5a, 0x016b5a, PG_U_UNASSIGNED},
{0x016b5b, 0x016b61, PG_U_OTHER_NUMBER},
- {0x016b62, 0x016b62, PG_U_UNASSIGNED},
{0x016b63, 0x016b77, PG_U_OTHER_LETTER},
- {0x016b78, 0x016b7c, PG_U_UNASSIGNED},
{0x016b7d, 0x016b8f, PG_U_OTHER_LETTER},
- {0x016b90, 0x016e3f, PG_U_UNASSIGNED},
{0x016e40, 0x016e5f, PG_U_UPPERCASE_LETTER},
{0x016e60, 0x016e7f, PG_U_LOWERCASE_LETTER},
{0x016e80, 0x016e96, PG_U_OTHER_NUMBER},
{0x016e97, 0x016e9a, PG_U_OTHER_PUNCTUATION},
- {0x016e9b, 0x016eff, PG_U_UNASSIGNED},
{0x016f00, 0x016f4a, PG_U_OTHER_LETTER},
- {0x016f4b, 0x016f4e, PG_U_UNASSIGNED},
{0x016f4f, 0x016f4f, PG_U_NONSPACING_MARK},
{0x016f50, 0x016f50, PG_U_OTHER_LETTER},
{0x016f51, 0x016f87, PG_U_SPACING_MARK},
- {0x016f88, 0x016f8e, PG_U_UNASSIGNED},
{0x016f8f, 0x016f92, PG_U_NONSPACING_MARK},
{0x016f93, 0x016f9f, PG_U_MODIFIER_LETTER},
- {0x016fa0, 0x016fdf, PG_U_UNASSIGNED},
{0x016fe0, 0x016fe1, PG_U_MODIFIER_LETTER},
{0x016fe2, 0x016fe2, PG_U_OTHER_PUNCTUATION},
{0x016fe3, 0x016fe3, PG_U_MODIFIER_LETTER},
{0x016fe4, 0x016fe4, PG_U_NONSPACING_MARK},
- {0x016fe5, 0x016fef, PG_U_UNASSIGNED},
{0x016ff0, 0x016ff1, PG_U_SPACING_MARK},
- {0x016ff2, 0x016fff, PG_U_UNASSIGNED},
{0x017000, 0x0187f7, PG_U_OTHER_LETTER},
- {0x0187f8, 0x0187ff, PG_U_UNASSIGNED},
{0x018800, 0x018cd5, PG_U_OTHER_LETTER},
- {0x018cd6, 0x018cff, PG_U_UNASSIGNED},
{0x018d00, 0x018d08, PG_U_OTHER_LETTER},
- {0x018d09, 0x01afef, PG_U_UNASSIGNED},
{0x01aff0, 0x01aff3, PG_U_MODIFIER_LETTER},
- {0x01aff4, 0x01aff4, PG_U_UNASSIGNED},
{0x01aff5, 0x01affb, PG_U_MODIFIER_LETTER},
- {0x01affc, 0x01affc, PG_U_UNASSIGNED},
{0x01affd, 0x01affe, PG_U_MODIFIER_LETTER},
- {0x01afff, 0x01afff, PG_U_UNASSIGNED},
{0x01b000, 0x01b122, PG_U_OTHER_LETTER},
- {0x01b123, 0x01b131, PG_U_UNASSIGNED},
{0x01b132, 0x01b132, PG_U_OTHER_LETTER},
- {0x01b133, 0x01b14f, PG_U_UNASSIGNED},
{0x01b150, 0x01b152, PG_U_OTHER_LETTER},
- {0x01b153, 0x01b154, PG_U_UNASSIGNED},
{0x01b155, 0x01b155, PG_U_OTHER_LETTER},
- {0x01b156, 0x01b163, PG_U_UNASSIGNED},
{0x01b164, 0x01b167, PG_U_OTHER_LETTER},
- {0x01b168, 0x01b16f, PG_U_UNASSIGNED},
{0x01b170, 0x01b2fb, PG_U_OTHER_LETTER},
- {0x01b2fc, 0x01bbff, PG_U_UNASSIGNED},
{0x01bc00, 0x01bc6a, PG_U_OTHER_LETTER},
- {0x01bc6b, 0x01bc6f, PG_U_UNASSIGNED},
{0x01bc70, 0x01bc7c, PG_U_OTHER_LETTER},
- {0x01bc7d, 0x01bc7f, PG_U_UNASSIGNED},
{0x01bc80, 0x01bc88, PG_U_OTHER_LETTER},
- {0x01bc89, 0x01bc8f, PG_U_UNASSIGNED},
{0x01bc90, 0x01bc99, PG_U_OTHER_LETTER},
- {0x01bc9a, 0x01bc9b, PG_U_UNASSIGNED},
{0x01bc9c, 0x01bc9c, PG_U_OTHER_SYMBOL},
{0x01bc9d, 0x01bc9e, PG_U_NONSPACING_MARK},
{0x01bc9f, 0x01bc9f, PG_U_OTHER_PUNCTUATION},
{0x01bca0, 0x01bca3, PG_U_FORMAT},
- {0x01bca4, 0x01ceff, PG_U_UNASSIGNED},
{0x01cf00, 0x01cf2d, PG_U_NONSPACING_MARK},
- {0x01cf2e, 0x01cf2f, PG_U_UNASSIGNED},
{0x01cf30, 0x01cf46, PG_U_NONSPACING_MARK},
- {0x01cf47, 0x01cf4f, PG_U_UNASSIGNED},
{0x01cf50, 0x01cfc3, PG_U_OTHER_SYMBOL},
- {0x01cfc4, 0x01cfff, PG_U_UNASSIGNED},
{0x01d000, 0x01d0f5, PG_U_OTHER_SYMBOL},
- {0x01d0f6, 0x01d0ff, PG_U_UNASSIGNED},
{0x01d100, 0x01d126, PG_U_OTHER_SYMBOL},
- {0x01d127, 0x01d128, PG_U_UNASSIGNED},
{0x01d129, 0x01d164, PG_U_OTHER_SYMBOL},
{0x01d165, 0x01d166, PG_U_SPACING_MARK},
{0x01d167, 0x01d169, PG_U_NONSPACING_MARK},
@@ -3670,66 +3105,42 @@ static const pg_category_range unicode_categories[4009] =
{0x01d18c, 0x01d1a9, PG_U_OTHER_SYMBOL},
{0x01d1aa, 0x01d1ad, PG_U_NONSPACING_MARK},
{0x01d1ae, 0x01d1ea, PG_U_OTHER_SYMBOL},
- {0x01d1eb, 0x01d1ff, PG_U_UNASSIGNED},
{0x01d200, 0x01d241, PG_U_OTHER_SYMBOL},
{0x01d242, 0x01d244, PG_U_NONSPACING_MARK},
{0x01d245, 0x01d245, PG_U_OTHER_SYMBOL},
- {0x01d246, 0x01d2bf, PG_U_UNASSIGNED},
{0x01d2c0, 0x01d2d3, PG_U_OTHER_NUMBER},
- {0x01d2d4, 0x01d2df, PG_U_UNASSIGNED},
{0x01d2e0, 0x01d2f3, PG_U_OTHER_NUMBER},
- {0x01d2f4, 0x01d2ff, PG_U_UNASSIGNED},
{0x01d300, 0x01d356, PG_U_OTHER_SYMBOL},
- {0x01d357, 0x01d35f, PG_U_UNASSIGNED},
{0x01d360, 0x01d378, PG_U_OTHER_NUMBER},
- {0x01d379, 0x01d3ff, PG_U_UNASSIGNED},
{0x01d400, 0x01d419, PG_U_UPPERCASE_LETTER},
{0x01d41a, 0x01d433, PG_U_LOWERCASE_LETTER},
{0x01d434, 0x01d44d, PG_U_UPPERCASE_LETTER},
{0x01d44e, 0x01d454, PG_U_LOWERCASE_LETTER},
- {0x01d455, 0x01d455, PG_U_UNASSIGNED},
{0x01d456, 0x01d467, PG_U_LOWERCASE_LETTER},
{0x01d468, 0x01d481, PG_U_UPPERCASE_LETTER},
{0x01d482, 0x01d49b, PG_U_LOWERCASE_LETTER},
{0x01d49c, 0x01d49c, PG_U_UPPERCASE_LETTER},
- {0x01d49d, 0x01d49d, PG_U_UNASSIGNED},
{0x01d49e, 0x01d49f, PG_U_UPPERCASE_LETTER},
- {0x01d4a0, 0x01d4a1, PG_U_UNASSIGNED},
{0x01d4a2, 0x01d4a2, PG_U_UPPERCASE_LETTER},
- {0x01d4a3, 0x01d4a4, PG_U_UNASSIGNED},
{0x01d4a5, 0x01d4a6, PG_U_UPPERCASE_LETTER},
- {0x01d4a7, 0x01d4a8, PG_U_UNASSIGNED},
{0x01d4a9, 0x01d4ac, PG_U_UPPERCASE_LETTER},
- {0x01d4ad, 0x01d4ad, PG_U_UNASSIGNED},
{0x01d4ae, 0x01d4b5, PG_U_UPPERCASE_LETTER},
{0x01d4b6, 0x01d4b9, PG_U_LOWERCASE_LETTER},
- {0x01d4ba, 0x01d4ba, PG_U_UNASSIGNED},
{0x01d4bb, 0x01d4bb, PG_U_LOWERCASE_LETTER},
- {0x01d4bc, 0x01d4bc, PG_U_UNASSIGNED},
{0x01d4bd, 0x01d4c3, PG_U_LOWERCASE_LETTER},
- {0x01d4c4, 0x01d4c4, PG_U_UNASSIGNED},
{0x01d4c5, 0x01d4cf, PG_U_LOWERCASE_LETTER},
{0x01d4d0, 0x01d4e9, PG_U_UPPERCASE_LETTER},
{0x01d4ea, 0x01d503, PG_U_LOWERCASE_LETTER},
{0x01d504, 0x01d505, PG_U_UPPERCASE_LETTER},
- {0x01d506, 0x01d506, PG_U_UNASSIGNED},
{0x01d507, 0x01d50a, PG_U_UPPERCASE_LETTER},
- {0x01d50b, 0x01d50c, PG_U_UNASSIGNED},
{0x01d50d, 0x01d514, PG_U_UPPERCASE_LETTER},
- {0x01d515, 0x01d515, PG_U_UNASSIGNED},
{0x01d516, 0x01d51c, PG_U_UPPERCASE_LETTER},
- {0x01d51d, 0x01d51d, PG_U_UNASSIGNED},
{0x01d51e, 0x01d537, PG_U_LOWERCASE_LETTER},
{0x01d538, 0x01d539, PG_U_UPPERCASE_LETTER},
- {0x01d53a, 0x01d53a, PG_U_UNASSIGNED},
{0x01d53b, 0x01d53e, PG_U_UPPERCASE_LETTER},
- {0x01d53f, 0x01d53f, PG_U_UNASSIGNED},
{0x01d540, 0x01d544, PG_U_UPPERCASE_LETTER},
- {0x01d545, 0x01d545, PG_U_UNASSIGNED},
{0x01d546, 0x01d546, PG_U_UPPERCASE_LETTER},
- {0x01d547, 0x01d549, PG_U_UNASSIGNED},
{0x01d54a, 0x01d550, PG_U_UPPERCASE_LETTER},
- {0x01d551, 0x01d551, PG_U_UNASSIGNED},
{0x01d552, 0x01d56b, PG_U_LOWERCASE_LETTER},
{0x01d56c, 0x01d585, PG_U_UPPERCASE_LETTER},
{0x01d586, 0x01d59f, PG_U_LOWERCASE_LETTER},
@@ -3743,7 +3154,6 @@ static const pg_category_range unicode_categories[4009] =
{0x01d656, 0x01d66f, PG_U_LOWERCASE_LETTER},
{0x01d670, 0x01d689, PG_U_UPPERCASE_LETTER},
{0x01d68a, 0x01d6a5, PG_U_LOWERCASE_LETTER},
- {0x01d6a6, 0x01d6a7, PG_U_UNASSIGNED},
{0x01d6a8, 0x01d6c0, PG_U_UPPERCASE_LETTER},
{0x01d6c1, 0x01d6c1, PG_U_MATH_SYMBOL},
{0x01d6c2, 0x01d6da, PG_U_LOWERCASE_LETTER},
@@ -3771,7 +3181,6 @@ static const pg_category_range unicode_categories[4009] =
{0x01d7c4, 0x01d7c9, PG_U_LOWERCASE_LETTER},
{0x01d7ca, 0x01d7ca, PG_U_UPPERCASE_LETTER},
{0x01d7cb, 0x01d7cb, PG_U_LOWERCASE_LETTER},
- {0x01d7cc, 0x01d7cd, PG_U_UNASSIGNED},
{0x01d7ce, 0x01d7ff, PG_U_DECIMAL_NUMBER},
{0x01d800, 0x01d9ff, PG_U_OTHER_SYMBOL},
{0x01da00, 0x01da36, PG_U_NONSPACING_MARK},
@@ -3783,258 +3192,142 @@ static const pg_category_range unicode_categories[4009] =
{0x01da84, 0x01da84, PG_U_NONSPACING_MARK},
{0x01da85, 0x01da86, PG_U_OTHER_SYMBOL},
{0x01da87, 0x01da8b, PG_U_OTHER_PUNCTUATION},
- {0x01da8c, 0x01da9a, PG_U_UNASSIGNED},
{0x01da9b, 0x01da9f, PG_U_NONSPACING_MARK},
- {0x01daa0, 0x01daa0, PG_U_UNASSIGNED},
{0x01daa1, 0x01daaf, PG_U_NONSPACING_MARK},
- {0x01dab0, 0x01deff, PG_U_UNASSIGNED},
{0x01df00, 0x01df09, PG_U_LOWERCASE_LETTER},
{0x01df0a, 0x01df0a, PG_U_OTHER_LETTER},
{0x01df0b, 0x01df1e, PG_U_LOWERCASE_LETTER},
- {0x01df1f, 0x01df24, PG_U_UNASSIGNED},
{0x01df25, 0x01df2a, PG_U_LOWERCASE_LETTER},
- {0x01df2b, 0x01dfff, PG_U_UNASSIGNED},
{0x01e000, 0x01e006, PG_U_NONSPACING_MARK},
- {0x01e007, 0x01e007, PG_U_UNASSIGNED},
{0x01e008, 0x01e018, PG_U_NONSPACING_MARK},
- {0x01e019, 0x01e01a, PG_U_UNASSIGNED},
{0x01e01b, 0x01e021, PG_U_NONSPACING_MARK},
- {0x01e022, 0x01e022, PG_U_UNASSIGNED},
{0x01e023, 0x01e024, PG_U_NONSPACING_MARK},
- {0x01e025, 0x01e025, PG_U_UNASSIGNED},
{0x01e026, 0x01e02a, PG_U_NONSPACING_MARK},
- {0x01e02b, 0x01e02f, PG_U_UNASSIGNED},
{0x01e030, 0x01e06d, PG_U_MODIFIER_LETTER},
- {0x01e06e, 0x01e08e, PG_U_UNASSIGNED},
{0x01e08f, 0x01e08f, PG_U_NONSPACING_MARK},
- {0x01e090, 0x01e0ff, PG_U_UNASSIGNED},
{0x01e100, 0x01e12c, PG_U_OTHER_LETTER},
- {0x01e12d, 0x01e12f, PG_U_UNASSIGNED},
{0x01e130, 0x01e136, PG_U_NONSPACING_MARK},
{0x01e137, 0x01e13d, PG_U_MODIFIER_LETTER},
- {0x01e13e, 0x01e13f, PG_U_UNASSIGNED},
{0x01e140, 0x01e149, PG_U_DECIMAL_NUMBER},
- {0x01e14a, 0x01e14d, PG_U_UNASSIGNED},
{0x01e14e, 0x01e14e, PG_U_OTHER_LETTER},
{0x01e14f, 0x01e14f, PG_U_OTHER_SYMBOL},
- {0x01e150, 0x01e28f, PG_U_UNASSIGNED},
{0x01e290, 0x01e2ad, PG_U_OTHER_LETTER},
{0x01e2ae, 0x01e2ae, PG_U_NONSPACING_MARK},
- {0x01e2af, 0x01e2bf, PG_U_UNASSIGNED},
{0x01e2c0, 0x01e2eb, PG_U_OTHER_LETTER},
{0x01e2ec, 0x01e2ef, PG_U_NONSPACING_MARK},
{0x01e2f0, 0x01e2f9, PG_U_DECIMAL_NUMBER},
- {0x01e2fa, 0x01e2fe, PG_U_UNASSIGNED},
{0x01e2ff, 0x01e2ff, PG_U_CURRENCY_SYMBOL},
- {0x01e300, 0x01e4cf, PG_U_UNASSIGNED},
{0x01e4d0, 0x01e4ea, PG_U_OTHER_LETTER},
{0x01e4eb, 0x01e4eb, PG_U_MODIFIER_LETTER},
{0x01e4ec, 0x01e4ef, PG_U_NONSPACING_MARK},
{0x01e4f0, 0x01e4f9, PG_U_DECIMAL_NUMBER},
- {0x01e4fa, 0x01e7df, PG_U_UNASSIGNED},
{0x01e7e0, 0x01e7e6, PG_U_OTHER_LETTER},
- {0x01e7e7, 0x01e7e7, PG_U_UNASSIGNED},
{0x01e7e8, 0x01e7eb, PG_U_OTHER_LETTER},
- {0x01e7ec, 0x01e7ec, PG_U_UNASSIGNED},
{0x01e7ed, 0x01e7ee, PG_U_OTHER_LETTER},
- {0x01e7ef, 0x01e7ef, PG_U_UNASSIGNED},
{0x01e7f0, 0x01e7fe, PG_U_OTHER_LETTER},
- {0x01e7ff, 0x01e7ff, PG_U_UNASSIGNED},
{0x01e800, 0x01e8c4, PG_U_OTHER_LETTER},
- {0x01e8c5, 0x01e8c6, PG_U_UNASSIGNED},
{0x01e8c7, 0x01e8cf, PG_U_OTHER_NUMBER},
{0x01e8d0, 0x01e8d6, PG_U_NONSPACING_MARK},
- {0x01e8d7, 0x01e8ff, PG_U_UNASSIGNED},
{0x01e900, 0x01e921, PG_U_UPPERCASE_LETTER},
{0x01e922, 0x01e943, PG_U_LOWERCASE_LETTER},
{0x01e944, 0x01e94a, PG_U_NONSPACING_MARK},
{0x01e94b, 0x01e94b, PG_U_MODIFIER_LETTER},
- {0x01e94c, 0x01e94f, PG_U_UNASSIGNED},
{0x01e950, 0x01e959, PG_U_DECIMAL_NUMBER},
- {0x01e95a, 0x01e95d, PG_U_UNASSIGNED},
{0x01e95e, 0x01e95f, PG_U_OTHER_PUNCTUATION},
- {0x01e960, 0x01ec70, PG_U_UNASSIGNED},
{0x01ec71, 0x01ecab, PG_U_OTHER_NUMBER},
{0x01ecac, 0x01ecac, PG_U_OTHER_SYMBOL},
{0x01ecad, 0x01ecaf, PG_U_OTHER_NUMBER},
{0x01ecb0, 0x01ecb0, PG_U_CURRENCY_SYMBOL},
{0x01ecb1, 0x01ecb4, PG_U_OTHER_NUMBER},
- {0x01ecb5, 0x01ed00, PG_U_UNASSIGNED},
{0x01ed01, 0x01ed2d, PG_U_OTHER_NUMBER},
{0x01ed2e, 0x01ed2e, PG_U_OTHER_SYMBOL},
{0x01ed2f, 0x01ed3d, PG_U_OTHER_NUMBER},
- {0x01ed3e, 0x01edff, PG_U_UNASSIGNED},
{0x01ee00, 0x01ee03, PG_U_OTHER_LETTER},
- {0x01ee04, 0x01ee04, PG_U_UNASSIGNED},
{0x01ee05, 0x01ee1f, PG_U_OTHER_LETTER},
- {0x01ee20, 0x01ee20, PG_U_UNASSIGNED},
{0x01ee21, 0x01ee22, PG_U_OTHER_LETTER},
- {0x01ee23, 0x01ee23, PG_U_UNASSIGNED},
{0x01ee24, 0x01ee24, PG_U_OTHER_LETTER},
- {0x01ee25, 0x01ee26, PG_U_UNASSIGNED},
{0x01ee27, 0x01ee27, PG_U_OTHER_LETTER},
- {0x01ee28, 0x01ee28, PG_U_UNASSIGNED},
{0x01ee29, 0x01ee32, PG_U_OTHER_LETTER},
- {0x01ee33, 0x01ee33, PG_U_UNASSIGNED},
{0x01ee34, 0x01ee37, PG_U_OTHER_LETTER},
- {0x01ee38, 0x01ee38, PG_U_UNASSIGNED},
{0x01ee39, 0x01ee39, PG_U_OTHER_LETTER},
- {0x01ee3a, 0x01ee3a, PG_U_UNASSIGNED},
{0x01ee3b, 0x01ee3b, PG_U_OTHER_LETTER},
- {0x01ee3c, 0x01ee41, PG_U_UNASSIGNED},
{0x01ee42, 0x01ee42, PG_U_OTHER_LETTER},
- {0x01ee43, 0x01ee46, PG_U_UNASSIGNED},
{0x01ee47, 0x01ee47, PG_U_OTHER_LETTER},
- {0x01ee48, 0x01ee48, PG_U_UNASSIGNED},
{0x01ee49, 0x01ee49, PG_U_OTHER_LETTER},
- {0x01ee4a, 0x01ee4a, PG_U_UNASSIGNED},
{0x01ee4b, 0x01ee4b, PG_U_OTHER_LETTER},
- {0x01ee4c, 0x01ee4c, PG_U_UNASSIGNED},
{0x01ee4d, 0x01ee4f, PG_U_OTHER_LETTER},
- {0x01ee50, 0x01ee50, PG_U_UNASSIGNED},
{0x01ee51, 0x01ee52, PG_U_OTHER_LETTER},
- {0x01ee53, 0x01ee53, PG_U_UNASSIGNED},
{0x01ee54, 0x01ee54, PG_U_OTHER_LETTER},
- {0x01ee55, 0x01ee56, PG_U_UNASSIGNED},
{0x01ee57, 0x01ee57, PG_U_OTHER_LETTER},
- {0x01ee58, 0x01ee58, PG_U_UNASSIGNED},
{0x01ee59, 0x01ee59, PG_U_OTHER_LETTER},
- {0x01ee5a, 0x01ee5a, PG_U_UNASSIGNED},
{0x01ee5b, 0x01ee5b, PG_U_OTHER_LETTER},
- {0x01ee5c, 0x01ee5c, PG_U_UNASSIGNED},
{0x01ee5d, 0x01ee5d, PG_U_OTHER_LETTER},
- {0x01ee5e, 0x01ee5e, PG_U_UNASSIGNED},
{0x01ee5f, 0x01ee5f, PG_U_OTHER_LETTER},
- {0x01ee60, 0x01ee60, PG_U_UNASSIGNED},
{0x01ee61, 0x01ee62, PG_U_OTHER_LETTER},
- {0x01ee63, 0x01ee63, PG_U_UNASSIGNED},
{0x01ee64, 0x01ee64, PG_U_OTHER_LETTER},
- {0x01ee65, 0x01ee66, PG_U_UNASSIGNED},
{0x01ee67, 0x01ee6a, PG_U_OTHER_LETTER},
- {0x01ee6b, 0x01ee6b, PG_U_UNASSIGNED},
{0x01ee6c, 0x01ee72, PG_U_OTHER_LETTER},
- {0x01ee73, 0x01ee73, PG_U_UNASSIGNED},
{0x01ee74, 0x01ee77, PG_U_OTHER_LETTER},
- {0x01ee78, 0x01ee78, PG_U_UNASSIGNED},
{0x01ee79, 0x01ee7c, PG_U_OTHER_LETTER},
- {0x01ee7d, 0x01ee7d, PG_U_UNASSIGNED},
{0x01ee7e, 0x01ee7e, PG_U_OTHER_LETTER},
- {0x01ee7f, 0x01ee7f, PG_U_UNASSIGNED},
{0x01ee80, 0x01ee89, PG_U_OTHER_LETTER},
- {0x01ee8a, 0x01ee8a, PG_U_UNASSIGNED},
{0x01ee8b, 0x01ee9b, PG_U_OTHER_LETTER},
- {0x01ee9c, 0x01eea0, PG_U_UNASSIGNED},
{0x01eea1, 0x01eea3, PG_U_OTHER_LETTER},
- {0x01eea4, 0x01eea4, PG_U_UNASSIGNED},
{0x01eea5, 0x01eea9, PG_U_OTHER_LETTER},
- {0x01eeaa, 0x01eeaa, PG_U_UNASSIGNED},
{0x01eeab, 0x01eebb, PG_U_OTHER_LETTER},
- {0x01eebc, 0x01eeef, PG_U_UNASSIGNED},
{0x01eef0, 0x01eef1, PG_U_MATH_SYMBOL},
- {0x01eef2, 0x01efff, PG_U_UNASSIGNED},
{0x01f000, 0x01f02b, PG_U_OTHER_SYMBOL},
- {0x01f02c, 0x01f02f, PG_U_UNASSIGNED},
{0x01f030, 0x01f093, PG_U_OTHER_SYMBOL},
- {0x01f094, 0x01f09f, PG_U_UNASSIGNED},
{0x01f0a0, 0x01f0ae, PG_U_OTHER_SYMBOL},
- {0x01f0af, 0x01f0b0, PG_U_UNASSIGNED},
{0x01f0b1, 0x01f0bf, PG_U_OTHER_SYMBOL},
- {0x01f0c0, 0x01f0c0, PG_U_UNASSIGNED},
{0x01f0c1, 0x01f0cf, PG_U_OTHER_SYMBOL},
- {0x01f0d0, 0x01f0d0, PG_U_UNASSIGNED},
{0x01f0d1, 0x01f0f5, PG_U_OTHER_SYMBOL},
- {0x01f0f6, 0x01f0ff, PG_U_UNASSIGNED},
{0x01f100, 0x01f10c, PG_U_OTHER_NUMBER},
{0x01f10d, 0x01f1ad, PG_U_OTHER_SYMBOL},
- {0x01f1ae, 0x01f1e5, PG_U_UNASSIGNED},
{0x01f1e6, 0x01f202, PG_U_OTHER_SYMBOL},
- {0x01f203, 0x01f20f, PG_U_UNASSIGNED},
{0x01f210, 0x01f23b, PG_U_OTHER_SYMBOL},
- {0x01f23c, 0x01f23f, PG_U_UNASSIGNED},
{0x01f240, 0x01f248, PG_U_OTHER_SYMBOL},
- {0x01f249, 0x01f24f, PG_U_UNASSIGNED},
{0x01f250, 0x01f251, PG_U_OTHER_SYMBOL},
- {0x01f252, 0x01f25f, PG_U_UNASSIGNED},
{0x01f260, 0x01f265, PG_U_OTHER_SYMBOL},
- {0x01f266, 0x01f2ff, PG_U_UNASSIGNED},
{0x01f300, 0x01f3fa, PG_U_OTHER_SYMBOL},
{0x01f3fb, 0x01f3ff, PG_U_MODIFIER_SYMBOL},
{0x01f400, 0x01f6d7, PG_U_OTHER_SYMBOL},
- {0x01f6d8, 0x01f6db, PG_U_UNASSIGNED},
{0x01f6dc, 0x01f6ec, PG_U_OTHER_SYMBOL},
- {0x01f6ed, 0x01f6ef, PG_U_UNASSIGNED},
{0x01f6f0, 0x01f6fc, PG_U_OTHER_SYMBOL},
- {0x01f6fd, 0x01f6ff, PG_U_UNASSIGNED},
{0x01f700, 0x01f776, PG_U_OTHER_SYMBOL},
- {0x01f777, 0x01f77a, PG_U_UNASSIGNED},
{0x01f77b, 0x01f7d9, PG_U_OTHER_SYMBOL},
- {0x01f7da, 0x01f7df, PG_U_UNASSIGNED},
{0x01f7e0, 0x01f7eb, PG_U_OTHER_SYMBOL},
- {0x01f7ec, 0x01f7ef, PG_U_UNASSIGNED},
{0x01f7f0, 0x01f7f0, PG_U_OTHER_SYMBOL},
- {0x01f7f1, 0x01f7ff, PG_U_UNASSIGNED},
{0x01f800, 0x01f80b, PG_U_OTHER_SYMBOL},
- {0x01f80c, 0x01f80f, PG_U_UNASSIGNED},
{0x01f810, 0x01f847, PG_U_OTHER_SYMBOL},
- {0x01f848, 0x01f84f, PG_U_UNASSIGNED},
{0x01f850, 0x01f859, PG_U_OTHER_SYMBOL},
- {0x01f85a, 0x01f85f, PG_U_UNASSIGNED},
{0x01f860, 0x01f887, PG_U_OTHER_SYMBOL},
- {0x01f888, 0x01f88f, PG_U_UNASSIGNED},
{0x01f890, 0x01f8ad, PG_U_OTHER_SYMBOL},
- {0x01f8ae, 0x01f8af, PG_U_UNASSIGNED},
{0x01f8b0, 0x01f8b1, PG_U_OTHER_SYMBOL},
- {0x01f8b2, 0x01f8ff, PG_U_UNASSIGNED},
{0x01f900, 0x01fa53, PG_U_OTHER_SYMBOL},
- {0x01fa54, 0x01fa5f, PG_U_UNASSIGNED},
{0x01fa60, 0x01fa6d, PG_U_OTHER_SYMBOL},
- {0x01fa6e, 0x01fa6f, PG_U_UNASSIGNED},
{0x01fa70, 0x01fa7c, PG_U_OTHER_SYMBOL},
- {0x01fa7d, 0x01fa7f, PG_U_UNASSIGNED},
{0x01fa80, 0x01fa88, PG_U_OTHER_SYMBOL},
- {0x01fa89, 0x01fa8f, PG_U_UNASSIGNED},
{0x01fa90, 0x01fabd, PG_U_OTHER_SYMBOL},
- {0x01fabe, 0x01fabe, PG_U_UNASSIGNED},
{0x01fabf, 0x01fac5, PG_U_OTHER_SYMBOL},
- {0x01fac6, 0x01facd, PG_U_UNASSIGNED},
{0x01face, 0x01fadb, PG_U_OTHER_SYMBOL},
- {0x01fadc, 0x01fadf, PG_U_UNASSIGNED},
{0x01fae0, 0x01fae8, PG_U_OTHER_SYMBOL},
- {0x01fae9, 0x01faef, PG_U_UNASSIGNED},
{0x01faf0, 0x01faf8, PG_U_OTHER_SYMBOL},
- {0x01faf9, 0x01faff, PG_U_UNASSIGNED},
{0x01fb00, 0x01fb92, PG_U_OTHER_SYMBOL},
- {0x01fb93, 0x01fb93, PG_U_UNASSIGNED},
{0x01fb94, 0x01fbca, PG_U_OTHER_SYMBOL},
- {0x01fbcb, 0x01fbef, PG_U_UNASSIGNED},
{0x01fbf0, 0x01fbf9, PG_U_DECIMAL_NUMBER},
- {0x01fbfa, 0x01ffff, PG_U_UNASSIGNED},
{0x020000, 0x02a6df, PG_U_OTHER_LETTER},
- {0x02a6e0, 0x02a6ff, PG_U_UNASSIGNED},
{0x02a700, 0x02b739, PG_U_OTHER_LETTER},
- {0x02b73a, 0x02b73f, PG_U_UNASSIGNED},
{0x02b740, 0x02b81d, PG_U_OTHER_LETTER},
- {0x02b81e, 0x02b81f, PG_U_UNASSIGNED},
{0x02b820, 0x02cea1, PG_U_OTHER_LETTER},
- {0x02cea2, 0x02ceaf, PG_U_UNASSIGNED},
{0x02ceb0, 0x02ebe0, PG_U_OTHER_LETTER},
- {0x02ebe1, 0x02ebef, PG_U_UNASSIGNED},
{0x02ebf0, 0x02ee5d, PG_U_OTHER_LETTER},
- {0x02ee5e, 0x02f7ff, PG_U_UNASSIGNED},
{0x02f800, 0x02fa1d, PG_U_OTHER_LETTER},
- {0x02fa1e, 0x02ffff, PG_U_UNASSIGNED},
{0x030000, 0x03134a, PG_U_OTHER_LETTER},
- {0x03134b, 0x03134f, PG_U_UNASSIGNED},
{0x031350, 0x0323af, PG_U_OTHER_LETTER},
- {0x0323b0, 0x0e0000, PG_U_UNASSIGNED},
{0x0e0001, 0x0e0001, PG_U_FORMAT},
- {0x0e0002, 0x0e001f, PG_U_UNASSIGNED},
{0x0e0020, 0x0e007f, PG_U_FORMAT},
- {0x0e0080, 0x0e00ff, PG_U_UNASSIGNED},
{0x0e0100, 0x0e01ef, PG_U_NONSPACING_MARK},
- {0x0e01f0, 0x0effff, PG_U_UNASSIGNED},
{0x0f0000, 0x0ffffd, PG_U_PRIVATE_USE},
- {0x0ffffe, 0x0fffff, PG_U_UNASSIGNED},
- {0x100000, 0x10fffd, PG_U_PRIVATE_USE},
- {0x10fffe, 0x10ffff, PG_U_UNASSIGNED}
+ {0x100000, 0x10fffd, PG_U_PRIVATE_USE}
};
--
2.34.1
[text/x-patch] v1-0001-Minor-cleanup-for-unicode-update-build-and-test.patch (7.4K, ../../[email protected]/4-v1-0001-Minor-cleanup-for-unicode-update-build-and-test.patch)
download | inline diff:
From 80ed701721b2bc91f2346f013d58930cd1d325f5 Mon Sep 17 00:00:00 2001
From: Jeff Davis <[email protected]>
Date: Wed, 22 Nov 2023 10:38:46 -0800
Subject: [PATCH v1 1/3] Minor cleanup for unicode-update build and test.
---
src/common/unicode/Makefile | 6 ++--
src/common/unicode/category_test.c | 18 ++++++------
src/common/unicode/meson.build | 44 +++++++++++++++---------------
3 files changed, 34 insertions(+), 34 deletions(-)
diff --git a/src/common/unicode/Makefile b/src/common/unicode/Makefile
index 30cd75cc6a..04d81dd5cb 100644
--- a/src/common/unicode/Makefile
+++ b/src/common/unicode/Makefile
@@ -21,7 +21,7 @@ CPPFLAGS += $(ICU_CFLAGS)
# By default, do nothing.
all:
-update-unicode: unicode_category_table.h unicode_norm_table.h unicode_nonspacing_table.h unicode_east_asian_fw_table.h unicode_normprops_table.h unicode_norm_hashfunc.h unicode_version.h
+update-unicode: unicode_category_table.h unicode_east_asian_fw_table.h unicode_nonspacing_table.h unicode_norm_hashfunc.h unicode_norm_table.h unicode_normprops_table.h unicode_version.h
mv $^ $(top_srcdir)/src/include/common/
$(MAKE) category-check
$(MAKE) normalization-check
@@ -29,7 +29,7 @@ update-unicode: unicode_category_table.h unicode_norm_table.h unicode_nonspacing
# These files are part of the Unicode Character Database. Download
# them on demand. The dependency on Makefile.global is for
# UNICODE_VERSION.
-UnicodeData.txt EastAsianWidth.txt DerivedNormalizationProps.txt CompositionExclusions.txt NormalizationTest.txt: $(top_builddir)/src/Makefile.global
+CompositionExclusions.txt DerivedNormalizationProps.txt EastAsianWidth.txt NormalizationTest.txt UnicodeData.txt: $(top_builddir)/src/Makefile.global
$(DOWNLOAD) https://www.unicode.org/Public/$(UNICODE_VERSION)/ucd/$(@F)
unicode_version.h: generate-unicode_version.pl
@@ -82,4 +82,4 @@ clean:
rm -f $(OBJS) category_test category_test.o norm_test norm_test.o
distclean: clean
- rm -f UnicodeData.txt EastAsianWidth.txt CompositionExclusions.txt NormalizationTest.txt norm_test_table.h unicode_norm_table.h
+ rm -f CompositionExclusions.txt DerivedNormalizationProps.txt EastAsianWidth.txt NormalizationTest.txt UnicodeData.txt norm_test_table.h unicode_category_table.h unicode_norm_table.h
diff --git a/src/common/unicode/category_test.c b/src/common/unicode/category_test.c
index ba62716d45..d9ea806eb8 100644
--- a/src/common/unicode/category_test.c
+++ b/src/common/unicode/category_test.c
@@ -54,8 +54,8 @@ main(int argc, char **argv)
int pg_skipped_codepoints = 0;
int icu_skipped_codepoints = 0;
- printf("Postgres Unicode Version:\t%s\n", PG_UNICODE_VERSION);
- printf("ICU Unicode Version:\t\t%s\n", U_UNICODE_VERSION);
+ printf("category_test: Postgres Unicode version:\t%s\n", PG_UNICODE_VERSION);
+ printf("category_test: ICU Unicode version:\t\t%s\n", U_UNICODE_VERSION);
for (UChar32 code = 0; code <= 0x10ffff; code++)
{
@@ -79,11 +79,11 @@ main(int argc, char **argv)
icu_skipped_codepoints++;
else
{
- printf("FAILURE for codepoint %06x\n", code);
- printf("Postgres category: %02d %s %s\n", pg_category,
+ printf("category_test: FAILURE for codepoint 0x%06x\n", code);
+ printf("category_test: Postgres category: %02d %s %s\n", pg_category,
unicode_category_abbrev(pg_category),
unicode_category_string(pg_category));
- printf("ICU category: %02d %s %s\n", icu_category,
+ printf("category_test: ICU category: %02d %s %s\n", icu_category,
unicode_category_abbrev(icu_category),
unicode_category_string(icu_category));
printf("\n");
@@ -93,16 +93,16 @@ main(int argc, char **argv)
}
if (pg_skipped_codepoints > 0)
- printf("Skipped %d codepoints unassigned in Postgres due to Unicode version mismatch.\n",
+ printf("category_test: skipped %d codepoints unassigned in Postgres due to Unicode version mismatch\n",
pg_skipped_codepoints);
if (icu_skipped_codepoints > 0)
- printf("Skipped %d codepoints unassigned in ICU due to Unicode version mismatch.\n",
+ printf("category_test: skipped %d codepoints unassigned in ICU due to Unicode version mismatch\n",
icu_skipped_codepoints);
- printf("category_test: All tests successful!\n");
+ printf("category_test: success\n");
exit(0);
#else
- printf("ICU support required for test; skipping.\n");
+ printf("category_test: ICU support required for test; skipping\n");
exit(0);
#endif
}
diff --git a/src/common/unicode/meson.build b/src/common/unicode/meson.build
index 6af46122c4..e8cfdc1df4 100644
--- a/src/common/unicode/meson.build
+++ b/src/common/unicode/meson.build
@@ -11,7 +11,7 @@ endif
# These files are part of the Unicode Character Database. Download them on
# demand.
-foreach f : ['UnicodeData.txt', 'EastAsianWidth.txt', 'DerivedNormalizationProps.txt', 'CompositionExclusions.txt', 'NormalizationTest.txt']
+foreach f : ['CompositionExclusions.txt', 'DerivedNormalizationProps.txt', 'EastAsianWidth.txt', 'NormalizationTest.txt', 'UnicodeData.txt']
url = unicode_baseurl.format(UNICODE_VERSION, f)
target = custom_target(f,
output: f,
@@ -24,15 +24,6 @@ endforeach
update_unicode_targets = []
-update_unicode_targets += \
- custom_target('unicode_version.h',
- output: ['unicode_version.h'],
- command: [
- perl, files('generate-unicode_version.pl'),
- '--outdir', '@OUTDIR@', '--version', UNICODE_VERSION],
- build_by_default: false,
- )
-
update_unicode_targets += \
custom_target('unicode_category_table.h',
input: [unicode_data['UnicodeData.txt']],
@@ -44,14 +35,12 @@ update_unicode_targets += \
)
update_unicode_targets += \
- custom_target('unicode_norm_table.h',
- input: [unicode_data['UnicodeData.txt'], unicode_data['CompositionExclusions.txt']],
- output: ['unicode_norm_table.h', 'unicode_norm_hashfunc.h'],
- depend_files: perfect_hash_pm,
- command: [
- perl, files('generate-unicode_norm_table.pl'),
- '--outdir', '@OUTDIR@', '@INPUT@'],
+ custom_target('unicode_east_asian_fw_table.h',
+ input: [unicode_data['EastAsianWidth.txt']],
+ output: ['unicode_east_asian_fw_table.h'],
+ command: [perl, files('generate-unicode_east_asian_fw_table.pl'), '@INPUT@'],
build_by_default: false,
+ capture: true,
)
update_unicode_targets += \
@@ -65,12 +54,14 @@ update_unicode_targets += \
)
update_unicode_targets += \
- custom_target('unicode_east_asian_fw_table.h',
- input: [unicode_data['EastAsianWidth.txt']],
- output: ['unicode_east_asian_fw_table.h'],
- command: [perl, files('generate-unicode_east_asian_fw_table.pl'), '@INPUT@'],
+ custom_target('unicode_norm_table.h',
+ input: [unicode_data['UnicodeData.txt'], unicode_data['CompositionExclusions.txt']],
+ output: ['unicode_norm_table.h', 'unicode_norm_hashfunc.h'],
+ depend_files: perfect_hash_pm,
+ command: [
+ perl, files('generate-unicode_norm_table.pl'),
+ '--outdir', '@OUTDIR@', '@INPUT@'],
build_by_default: false,
- capture: true,
)
update_unicode_targets += \
@@ -83,6 +74,15 @@ update_unicode_targets += \
capture: true,
)
+update_unicode_targets += \
+ custom_target('unicode_version.h',
+ output: ['unicode_version.h'],
+ command: [
+ perl, files('generate-unicode_version.pl'),
+ '--outdir', '@OUTDIR@', '--version', UNICODE_VERSION],
+ build_by_default: false,
+ )
+
norm_test_table = custom_target('norm_test_table.h',
input: [unicode_data['NormalizationTest.txt']],
output: ['norm_test_table.h'],
--
2.34.1
^ permalink raw reply [nested|flat] 57+ messages in thread
* Re: encoding affects ICU regex character classification
2023-11-30 00:23 Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
@ 2023-11-30 02:10 ` Thomas Munro <[email protected]>
2023-12-01 20:49 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
0 siblings, 1 reply; 57+ messages in thread
From: Thomas Munro @ 2023-11-30 02:10 UTC (permalink / raw)
To: Jeff Davis <[email protected]>; +Cc: Tom Lane <[email protected]>; pgsql-hackers
On Thu, Nov 30, 2023 at 1:23 PM Jeff Davis <[email protected]> wrote:
> Character classification is not localized at all in libc or ICU as far
> as I can tell.
Really? POSIX isalpha()/isalpha_l() and friends clearly depend on a
locale. See eg d522b05c for a case where that broke something.
Perhaps you mean glibc wouldn't do that to you because you know that,
as an unstandardised detail, it sucks in (some version of) Unicode's
data which shouldn't vary between locales. But you are allowed to
make your own locales, including putting whatever classifications you
want into the LC_TYPE file using POSIX-standardised tools like
localedef. Perhaps that is a bit of a stretch, and no one really does
that in practice, but anyway it's still "localized".
Not knowing anything about how glibc generates its charmaps, Unicode
or pre-Unicode, I could take a wild guess that maybe in LATIN9 they
have an old hand-crafted table, but for UTF-8 encoding it's fully
outsourced to Unicode, and that's why you see a difference. Another
problem seen in a few parts of our tree is that we sometimes feed
individual UTF-8 bytes to the isXXX() functions which is about as well
defined as trying to pay for a pint with the left half of a $10 bill.
As for ICU, it's "not localized" only if there is only one ICU library
in the universe, but of course different versions of ICU might give
different answers because they correspond to different versions of
Unicode (as do glibc versions, FreeBSD libc versions, etc) and also
might disagree with tables built by PostgreSQL. Maybe irrelevant for
now, but I think with thus-far-imagined variants of the multi-version
ICU proposal, you have to choose whether to call u_isUAlphabetic() in
the library we're linked against, or via the dlsym() we look up in a
particular dlopen'd library. So I guess we'd have to access it via
our pg_locale_t, so again it'd be "localized" by some definitions.
Thinking about how to apply that thinking to libc, ... this is going
to sound far fetched and handwavy but here goes: we could even
imagine a multi-version system based on different base locale paths.
Instead of using the system-provided locales under /usr/share/locale
to look when we call newlocale(..., "en_NZ.UTF-8", ...), POSIX says
we're allowed to specify an absolute path eg newlocale(...,
"/foo/bar/unicode11/en_NZ.UTF-8", ...). If it is possible to use
$DISTRO's localedef to compile $OLD_DISTRO's locale sources to get
historical behaviour, that might provide a way to get them without
assuming the binary format is stable (it definitely isn't, but the
source format is nailed down by POSIX). One fly in the ointment is
that glibc failed to implement absolute path support, so you might
need to use versioned locale names instead, or see if the LOCPATH
environment variable can be swizzled around without confusing glibc's
locale cache. Then wouldn't be fundamentally different than the
hypothesised multi-version ICU case: you could probably come up with
different isalpha_l() results for different locales because you have
different LC_CTYPE versions (for example Unicode 15.0 added new
extended Cyrillic characters 1E030..1E08F, they look alphabetical to
me but what would I know). That is an extremely hypothetical
pie-in-the-sky thought and I don't know if it'd really work very well,
but it is a concrete way that someone might finish up getting
different answers out of isalpha_l(), to observe that it really is
localised. And localized.
^ permalink raw reply [nested|flat] 57+ messages in thread
* Re: encoding affects ICU regex character classification
2023-11-30 00:23 Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-11-30 02:10 ` Re: encoding affects ICU regex character classification Thomas Munro <[email protected]>
@ 2023-12-01 20:49 ` Jeff Davis <[email protected]>
2023-12-09 21:39 ` Re: encoding affects ICU regex character classification Thomas Munro <[email protected]>
0 siblings, 1 reply; 57+ messages in thread
From: Jeff Davis @ 2023-12-01 20:49 UTC (permalink / raw)
To: Thomas Munro <[email protected]>; +Cc: Tom Lane <[email protected]>; pgsql-hackers
On Thu, 2023-11-30 at 15:10 +1300, Thomas Munro wrote:
> > On Thu, Nov 30, 2023 at 1:23 PM Jeff Davis <[email protected]>
> > wrote:
> > > > Character classification is not localized at all in libc or ICU
> > > > as > > far
> > > > as I can tell.
> >
> > Really? POSIX isalpha()/isalpha_l() and friends clearly depend on
> > a
> > locale. See eg d522b05c for a case where that broke something.
I believe we're using different definitions of "localized". What I mean
is "varies from region to region or language to language". I think you
mean "varies for any reason at all [perhaps for no reason?]".
For instance, that commit indirectly links to:
https://github.com/evanj/isspace_locale
Which says "Mac OS X in a UTF-8 locale...". I don't see any fundamental
locale-based concern there.
I wrote a test program (attached) which compares any two given libc
locales using both the ordinary isalpha() family of functions, and also
using the iswalpha() family of functions. For the former, I only test
up to 0x7f. For the latter, I went to some effort to properly translate
the code point to a wchar_t (encode as UTF8, then mbstowcs using a UTF-
8 locale), and I test all unicode code points except the surrogate
range.
Using the test program, I compared the C.UTF-8 locale to every other
installed locale on my system (attached list for reference) and the
only ones that show any differences are "C" and "POSIX". That, combined
with the fact that ICU doesn't even accept a locale argument to the
character classification functions, gives me a high degree of
confidence that character classification is not localized on my system
according to my definition of "localized". If someone else wants to run
the test program on their system, I'd be interested to see the results
(some platform-specific modification may be required, e.g. handling 16-
bit whcar_t, etc.).
Your definition is too wide in my opinion, because it mixes together
different sources of variation that are best left separate:
a. region/language
b. technical requirements
c. versioning
d. implementation variance
(a) is not a true source of variation (please correct me if I'm wrong)
(b) is perhaps interesting. The "C" locale is one example, and perhaps
there are others, but I doubt very many others that we want to support.
(c) is not a major concern in my opinion. The impact of Unicode changes
is usually not dramatic, and it only affects regexes so it's much more
contained than collation, for example. And if you really care, just use
the "C" locale.
(d) is mostly a bug. Most users would prefer standardization, platform-
independence, documentability, and testability. There are users who
might care a lot about compatibility, and I don't want to disrupt such
users, but long term I don't see a lot of value in bubbling up
semantics from libc into expressions when there's not a clear reason to
do so. (Note: getting semantics from libc is a bit dubious in the case
of collation, as well, but at least for collation there are regional
and linguistic differences that we can't handle internally.)
I think we only need 2 main character classification schemes: "C" and
Unicode (TR #18 Compatibility Properties[1], either the "Standard"
variant or the "POSIX Compatible" variant or both). The libc and ICU
ones should be there only for compatibility and discouraged and
hopefully eventually removed.
> > Not knowing anything about how glibc generates its charmaps,
> > Unicode
> > or pre-Unicode, I could take a wild guess that maybe in LATIN9 they
> > have an old hand-crafted table, but for UTF-8 encoding it's fully
> > outsourced to Unicode, and that's why you see a difference.
No, the problem is that we're passing a pg_wchar to an ICU function
that expects a 32-bit code point. Those two things are equivalent in
the UTF8 encoding, but not in the LATIN9 encoding.
See the comment at the top of regc_pg_locale.c, which should probably
be updated to describe what happens with ICU collations.
> > Another
> > problem seen in a few parts of our tree is that we sometimes feed
> > individual UTF-8 bytes to the isXXX() functions which is about as >
> > well
> > defined as trying to pay for a pint with the left half of a $10
> > bill.
If we have built-in character classification systems as I propose ("C"
and Unicode), then the callers can simply choose which well-defined one
to use.
> > also
> > might disagree with tables built by PostgreSQL.
The patch I provided (new version attached) exhaustively tests all the
new Unicode property tables, and also the class assignments based on
[1] and [2]. Everything is identical for all assigned code points. The
test will run whenever you "ninja update-unicode", so any
inconsistencies will be highly visible before release. Additionally,
because the tables are checked in, you'll be able to see (in the diff)
the impact from a Unicode version update and consider that impact when
writing the release notes.
You may be wondering about differences in the version of Unicode
between Postgres and ICU while the test is running. It only tests code
points that are assigned in both Unicode versions, and reports the
number of code points that are skipped due to this check. The person
running "update-unicode" may see a failing test or a large number of
skipped codepoints if the Unicode versions don't match, in which case
they should try running against a more closely-matching version of ICU.
Regards,
Jeff Davis
[1] http://www.unicode.org/reports/tr18/#Compatibility_Properties
[2]
https://unicode-org.github.io/icu-docs/apidoc/dev/icu4c/uchar_8h.html#details
ar_AE
ar_AE.iso88596
ar_AE.utf8
C
cs_CZ.utf8
C.utf8
de_AT.utf8
de_BE.utf8
de_CH.utf8
de_DE.utf8
de_IT.utf8
de_LI.utf8
de_LU.utf8
en_AG
en_AG.utf8
en_AU.utf8
en_BW.utf8
en_CA.utf8
en_DK.utf8
en_GB.utf8
en_HK.utf8
en_IE.utf8
en_IL
en_IL.utf8
en_IN
en_IN.utf8
en_NG
en_NG.utf8
en_NZ.utf8
en_PH.utf8
en_SG.utf8
en_US
en_US.iso88591
en_US.iso885915
en_US.utf8
en_ZA.utf8
en_ZM
en_ZM.utf8
en_ZW.utf8
es_ES.utf8
es_MX.utf8
es_PE.utf8
es_US.utf8
fr_BE.utf8
fr_CA.utf8
fr_CH.utf8
french
fr_FR
fr_FR@euro
fr_FR.iso88591
fr_FR.iso885915@euro
fr_FR.utf8
fr_LU.utf8
ja_JP.utf8
ml_IN
ml_IN.utf8
POSIX
ru_RU.koi8r
russian
sv_SE.utf8
tr_TR
tr_TR.iso88599
tr_TR.utf8
turkish
zh_CN.utf8
zh_HK.utf8
zh_SG.utf8
zh_TW.utf8
Attachments:
[text/plain] l.txt (702B, ../../[email protected]/2-l.txt)
download | inline:
ar_AE
ar_AE.iso88596
ar_AE.utf8
C
cs_CZ.utf8
C.utf8
de_AT.utf8
de_BE.utf8
de_CH.utf8
de_DE.utf8
de_IT.utf8
de_LI.utf8
de_LU.utf8
en_AG
en_AG.utf8
en_AU.utf8
en_BW.utf8
en_CA.utf8
en_DK.utf8
en_GB.utf8
en_HK.utf8
en_IE.utf8
en_IL
en_IL.utf8
en_IN
en_IN.utf8
en_NG
en_NG.utf8
en_NZ.utf8
en_PH.utf8
en_SG.utf8
en_US
en_US.iso88591
en_US.iso885915
en_US.utf8
en_ZA.utf8
en_ZM
en_ZM.utf8
en_ZW.utf8
es_ES.utf8
es_MX.utf8
es_PE.utf8
es_US.utf8
fr_BE.utf8
fr_CA.utf8
fr_CH.utf8
french
fr_FR
fr_FR@euro
fr_FR.iso88591
fr_FR.iso885915@euro
fr_FR.utf8
fr_LU.utf8
ja_JP.utf8
ml_IN
ml_IN.utf8
POSIX
ru_RU.koi8r
russian
sv_SE.utf8
tr_TR
tr_TR.iso88599
tr_TR.utf8
turkish
zh_CN.utf8
zh_HK.utf8
zh_SG.utf8
zh_TW.utf8
[text/x-csrc] ctype_test.c (6.8K, ../../[email protected]/3-ctype_test.c)
download | inline:
#include <ctype.h>
#include <langinfo.h>
#include <locale.h>
#include <stdbool.h>
#include <stddef.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <wctype.h>
locale_t c_utf8_locale = NULL;
void
unicode_to_utf8(unsigned long int c, unsigned char *utf8string)
{
if (c <= 0x7F)
{
utf8string[0] = c;
}
else if (c <= 0x7FF)
{
utf8string[0] = 0xC0 | ((c >> 6) & 0x1F);
utf8string[1] = 0x80 | (c & 0x3F);
}
else if (c <= 0xFFFF)
{
utf8string[0] = 0xE0 | ((c >> 12) & 0x0F);
utf8string[1] = 0x80 | ((c >> 6) & 0x3F);
utf8string[2] = 0x80 | (c & 0x3F);
}
else
{
utf8string[0] = 0xF0 | ((c >> 18) & 0x07);
utf8string[1] = 0x80 | ((c >> 12) & 0x3F);
utf8string[2] = 0x80 | ((c >> 6) & 0x3F);
utf8string[3] = 0x80 | (c & 0x3F);
}
}
size_t
mbstowcs_l(wchar_t *dest, const char *src, size_t n, locale_t loc)
{
locale_t save_locale = uselocale(loc);
size_t result;
result = mbstowcs(dest, src, n);
uselocale(save_locale);
return result;
}
wchar_t
unicode_to_wchar_l(int codepoint, locale_t locale)
{
unsigned char utf8_string[5] = {0};
wchar_t wc[3] = {0};
size_t n;
if (codepoint == 0x000000)
return 0x00;
if (!c_utf8_locale)
{
c_utf8_locale = newlocale(LC_CTYPE_MASK, "C.UTF-8", NULL);
if (!c_utf8_locale)
{
printf("locale \"C.UTF-8\" not available\n");
exit(1);
}
}
memset(utf8_string, 0, 5);
unicode_to_utf8(codepoint, utf8_string);
n = mbstowcs_l(NULL, utf8_string, 0, c_utf8_locale);
if (n == -1)
{
printf("could not convert 0x%06x to wchar_t: invalid byte sequence\n",
codepoint);
printf("utf8_string: 0x%02x 0x%02x 0x%02x 0x%02x 0x%02x\n",
utf8_string[0], utf8_string[1], utf8_string[2], utf8_string[3], utf8_string[4]);
exit(1);
}
mbstowcs_l(wc, utf8_string, 2, c_utf8_locale);
return wc[0];
}
int main(int argc, char *argv[])
{
char *base_locale_string, *test_locale_string;
locale_t base_locale, test_locale;
if (argc < 2)
{
printf("Usage: ctype_test BASE_LOCALE TEST_LOCALE\n");
exit(1);
}
base_locale_string = argv[1];
test_locale_string = argv[2];
base_locale = newlocale(LC_CTYPE_MASK, base_locale_string, NULL);
if (!base_locale)
{
printf("base locale \"%s\" not available\n", base_locale_string);
exit(1);
}
test_locale = newlocale(LC_CTYPE_MASK, test_locale_string, NULL);
if (!test_locale)
{
printf("test locale \"%s\" not available\n", test_locale_string);
exit(1);
}
/* test plain char variants */
for (int i = -1; i <= 0x7f; i++)
{
bool base_isalpha = isalpha_l(i, base_locale);
bool base_islower = islower_l(i, base_locale);
bool base_isupper = isupper_l(i, base_locale);
bool base_isdigit = isdigit_l(i, base_locale);
bool base_isxdigit = isxdigit_l(i, base_locale);
bool base_isalnum = isalnum_l(i, base_locale);
bool base_isspace = isspace_l(i, base_locale);
bool base_ispunct = ispunct_l(i, base_locale);
bool base_isblank = isblank_l(i, base_locale);
bool base_isprint = isprint_l(i, base_locale);
bool base_isgraph = isgraph_l(i, base_locale);
bool base_iscntrl = iscntrl_l(i, base_locale);
bool test_isalpha = isalpha_l(i, test_locale);
bool test_islower = islower_l(i, test_locale);
bool test_isupper = isupper_l(i, test_locale);
bool test_isdigit = isdigit_l(i, test_locale);
bool test_isxdigit = isxdigit_l(i, test_locale);
bool test_isalnum = isalnum_l(i, test_locale);
bool test_isspace = isspace_l(i, test_locale);
bool test_ispunct = ispunct_l(i, test_locale);
bool test_isblank = isblank_l(i, test_locale);
bool test_isprint = isprint_l(i, test_locale);
bool test_isgraph = isgraph_l(i, test_locale);
bool test_iscntrl = iscntrl_l(i, test_locale);
if (base_isalpha != test_isalpha ||
base_islower != test_islower ||
base_isupper != test_isupper ||
base_isdigit != test_isdigit ||
base_isxdigit != test_isxdigit ||
base_isalnum != test_isalnum ||
base_isspace != test_isspace ||
base_ispunct != test_ispunct ||
base_isblank != test_isblank ||
base_isprint != test_isprint ||
base_isgraph != test_isgraph ||
base_iscntrl != test_iscntrl)
{
printf("FAILURE (%s,%s) for plain char 0x%02x\n",
base_locale_string, test_locale_string, i);
exit(1);
}
}
/* test wide variants */
for (wchar_t unicode = 0; unicode < 0x10ffff; unicode++)
{
/* skip surrogates */
if (unicode >= 0xd800 && unicode < 0xE000)
continue;
wchar_t wc1 = unicode_to_wchar_l(unicode, base_locale);
wchar_t wc2 = unicode_to_wchar_l(unicode, test_locale);
bool base_isalpha = iswalpha_l(wc1, base_locale);
bool base_islower = iswlower_l(wc1, base_locale);
bool base_isupper = iswupper_l(wc1, base_locale);
bool base_isdigit = iswdigit_l(wc1, base_locale);
bool base_isxdigit = iswxdigit_l(wc1, base_locale);
bool base_isalnum = iswalnum_l(wc1, base_locale);
bool base_isspace = iswspace_l(wc1, base_locale);
bool base_ispunct = iswpunct_l(wc1, base_locale);
bool base_isblank = iswblank_l(wc1, base_locale);
bool base_isprint = iswprint_l(wc1, base_locale);
bool base_isgraph = iswgraph_l(wc1, base_locale);
bool base_iscntrl = iswcntrl_l(wc1, base_locale);
bool test_isalpha = iswalpha_l(wc2, test_locale);
bool test_islower = iswlower_l(wc2, test_locale);
bool test_isupper = iswupper_l(wc2, test_locale);
bool test_isdigit = iswdigit_l(wc2, test_locale);
bool test_isxdigit = iswxdigit_l(wc2, test_locale);
bool test_isalnum = iswalnum_l(wc2, test_locale);
bool test_isspace = iswspace_l(wc2, test_locale);
bool test_ispunct = iswpunct_l(wc2, test_locale);
bool test_isblank = iswblank_l(wc2, test_locale);
bool test_isprint = iswprint_l(wc2, test_locale);
bool test_isgraph = iswgraph_l(wc2, test_locale);
bool test_iscntrl = iswcntrl_l(wc2, test_locale);
/* just a sanity check that works on some platforms, but not guaranteed */
if (unicode != wc1 || unicode != wc2)
printf("codepoint 0x%06x: 0x%06x 0x%06x\n", unicode, wc1, wc2);
if (base_isalpha != test_isalpha ||
base_islower != test_islower ||
base_isupper != test_isupper ||
base_isdigit != test_isdigit ||
base_isxdigit != test_isxdigit ||
base_isalnum != test_isalnum ||
base_isspace != test_isspace ||
base_ispunct != test_ispunct ||
base_isblank != test_isblank ||
base_isprint != test_isprint ||
base_isgraph != test_isgraph ||
base_iscntrl != test_iscntrl)
{
printf("FAILURE (%s,%s) for unicode char 0x%06x\n",
base_locale_string, test_locale_string, unicode);
exit(1);
}
}
freelocale(base_locale);
freelocale(test_locale);
}
[text/x-patch] v2-0003-Add-Unicode-property-tables.patch (91.5K, ../../[email protected]/4-v2-0003-Add-Unicode-property-tables.patch)
download | inline diff:
From 31f8a02ad90d9d03ed39bb09f7a585e37e72b4e8 Mon Sep 17 00:00:00 2001
From: Jeff Davis <[email protected]>
Date: Sat, 18 Nov 2023 15:34:24 -0800
Subject: [PATCH v2 3/3] Add Unicode property tables.
---
src/common/unicode/Makefile | 6 +-
src/common/unicode/category_test.c | 296 +-
.../generate-unicode_category_table.pl | 203 +-
src/common/unicode/meson.build | 4 +-
src/common/unicode_category.c | 266 +-
src/include/common/unicode_category.h | 29 +-
src/include/common/unicode_category_table.h | 2532 +++++++++++++++++
7 files changed, 3273 insertions(+), 63 deletions(-)
diff --git a/src/common/unicode/Makefile b/src/common/unicode/Makefile
index 04d81dd5cb..27f0408d8b 100644
--- a/src/common/unicode/Makefile
+++ b/src/common/unicode/Makefile
@@ -29,13 +29,13 @@ update-unicode: unicode_category_table.h unicode_east_asian_fw_table.h unicode_n
# These files are part of the Unicode Character Database. Download
# them on demand. The dependency on Makefile.global is for
# UNICODE_VERSION.
-CompositionExclusions.txt DerivedNormalizationProps.txt EastAsianWidth.txt NormalizationTest.txt UnicodeData.txt: $(top_builddir)/src/Makefile.global
+CompositionExclusions.txt DerivedCoreProperties.txt DerivedNormalizationProps.txt EastAsianWidth.txt NormalizationTest.txt PropList.txt UnicodeData.txt: $(top_builddir)/src/Makefile.global
$(DOWNLOAD) https://www.unicode.org/Public/$(UNICODE_VERSION)/ucd/$(@F)
unicode_version.h: generate-unicode_version.pl
$(PERL) $< --version $(UNICODE_VERSION)
-unicode_category_table.h: generate-unicode_category_table.pl UnicodeData.txt
+unicode_category_table.h: generate-unicode_category_table.pl DerivedCoreProperties.txt PropList.txt UnicodeData.txt
$(PERL) $<
# Generation of conversion tables used for string normalization with
@@ -82,4 +82,4 @@ clean:
rm -f $(OBJS) category_test category_test.o norm_test norm_test.o
distclean: clean
- rm -f CompositionExclusions.txt DerivedNormalizationProps.txt EastAsianWidth.txt NormalizationTest.txt UnicodeData.txt norm_test_table.h unicode_category_table.h unicode_norm_table.h
+ rm -f CompositionExclusions.txt DerivedCoreProperties.txt DerivedNormalizationProps.txt EastAsianWidth.txt NormalizationTest.txt PropList.txt UnicodeData.txt norm_test_table.h unicode_category_table.h unicode_norm_table.h
diff --git a/src/common/unicode/category_test.c b/src/common/unicode/category_test.c
index d9ea806eb8..b769850bf7 100644
--- a/src/common/unicode/category_test.c
+++ b/src/common/unicode/category_test.c
@@ -1,6 +1,7 @@
/*-------------------------------------------------------------------------
* category_test.c
- * Program to test Unicode general category functions.
+ * Program to test Unicode general category and character class
+ * functions.
*
* Portions Copyright (c) 2017-2023, PostgreSQL Global Development Group
*
@@ -14,17 +15,24 @@
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
-
#ifdef USE_ICU
#include <unicode/uchar.h>
#endif
-#include "common/unicode_category.h"
+#include <wctype.h>
+
#include "common/unicode_version.h"
+#include "common/unicode_category.h"
+
+#define LIBC_LOCALE "C.UTF-8"
+
+static int pg_unicode_version = 0;
+#ifdef USE_ICU
+static int icu_unicode_version = 0;
+#endif
/*
* Parse version into integer for easy comparison.
*/
-#ifdef USE_ICU
static int
parse_unicode_version(const char *version)
{
@@ -39,56 +47,160 @@ parse_unicode_version(const char *version)
return major * 100 + minor;
}
-#endif
+#ifdef USE_ICU
/*
- * Exhaustively test that the Unicode category for each codepoint matches that
- * returned by ICU.
+ * Test Postgres Unicode tables by comparing with ICU. Test the General
+ * Category, as well as the properties Alphabetic, Lowercase, Uppercase,
+ * White_Space, and Hex_Digit.
*/
-int
-main(int argc, char **argv)
+static void
+icu_test()
{
-#ifdef USE_ICU
- int pg_unicode_version = parse_unicode_version(PG_UNICODE_VERSION);
- int icu_unicode_version = parse_unicode_version(U_UNICODE_VERSION);
int pg_skipped_codepoints = 0;
int icu_skipped_codepoints = 0;
- printf("category_test: Postgres Unicode version:\t%s\n", PG_UNICODE_VERSION);
- printf("category_test: ICU Unicode version:\t\t%s\n", U_UNICODE_VERSION);
-
- for (UChar32 code = 0; code <= 0x10ffff; code++)
+ for (pg_wchar code = 0; code <= 0x10ffff; code++)
{
uint8_t pg_category = unicode_category(code);
uint8_t icu_category = u_charType(code);
+ /* Property tests */
+ bool prop_alphabetic = pg_u_prop_alphabetic(code);
+ bool prop_lowercase = pg_u_prop_lowercase(code);
+ bool prop_uppercase = pg_u_prop_uppercase(code);
+ bool prop_white_space = pg_u_prop_white_space(code);
+ bool prop_hex_digit = pg_u_prop_hex_digit(code);
+ bool prop_join_control = pg_u_prop_join_control(code);
+
+ bool icu_prop_alphabetic = u_hasBinaryProperty(
+ code, UCHAR_ALPHABETIC);
+ bool icu_prop_lowercase = u_hasBinaryProperty(
+ code, UCHAR_LOWERCASE);
+ bool icu_prop_uppercase = u_hasBinaryProperty(
+ code, UCHAR_UPPERCASE);
+ bool icu_prop_white_space = u_hasBinaryProperty(
+ code, UCHAR_WHITE_SPACE);
+ bool icu_prop_hex_digit = u_hasBinaryProperty(
+ code, UCHAR_HEX_DIGIT);
+ bool icu_prop_join_control = u_hasBinaryProperty(
+ code, UCHAR_JOIN_CONTROL);
+
+ /*
+ * Compare with ICU for TR #18 character classes using:
+ *
+ * https://unicode-org.github.io/icu-docs/apidoc/dev/icu4c/uchar_8h.html#details
+ *
+ * which describes how to use ICU to test for membership in regex
+ * character classes ("Standard", not "POSIX Compatible").
+ *
+ * NB: the document suggests testing for some properties such as
+ * UCHAR_POSIX_ALNUM, but that doesn't mean that we're testing for the
+ * "POSIX Compatible" character classes.
+ */
+ bool isalpha = pg_u_isalpha(code);
+ bool islower = pg_u_islower(code);
+ bool isupper = pg_u_isupper(code);
+ bool ispunct = pg_u_ispunct(code);
+ bool isdigit = pg_u_isdigit(code);
+ bool isxdigit = pg_u_isxdigit(code);
+ bool isalnum = pg_u_isalnum(code);
+ bool isspace = pg_u_isspace(code);
+ bool isblank = pg_u_isblank(code);
+ bool iscntrl = pg_u_iscntrl(code);
+ bool isgraph = pg_u_isgraph(code);
+ bool isprint = pg_u_isprint(code);
+
+ bool icu_isalpha = u_isUAlphabetic(code);
+ bool icu_islower = u_isULowercase(code);
+ bool icu_isupper = u_isUUppercase(code);
+ bool icu_ispunct = u_ispunct(code);
+ bool icu_isdigit = u_isdigit(code);
+ bool icu_isxdigit = u_hasBinaryProperty(code,
+ UCHAR_POSIX_XDIGIT);
+ bool icu_isalnum = u_hasBinaryProperty(code,
+ UCHAR_POSIX_ALNUM);
+ bool icu_isspace = u_isUWhiteSpace(code);
+ bool icu_isblank = u_isblank(code);
+ bool icu_iscntrl = icu_category == PG_U_CONTROL;
+ bool icu_isgraph = u_hasBinaryProperty(code,
+ UCHAR_POSIX_GRAPH);
+ bool icu_isprint = u_hasBinaryProperty(code,
+ UCHAR_POSIX_PRINT);
+
+ /*
+ * A version mismatch means that some assigned codepoints in the newer
+ * version may be unassigned in the older version. That's OK, though
+ * the test will not cover those codepoints marked unassigned in the
+ * older version (that is, it will no longer be an exhaustive test).
+ */
+ if (pg_category == PG_U_UNASSIGNED &&
+ icu_category != PG_U_UNASSIGNED &&
+ pg_unicode_version < icu_unicode_version)
+ {
+ pg_skipped_codepoints++;
+ continue;
+ }
+
+ if (icu_category == PG_U_UNASSIGNED &&
+ pg_category != PG_U_UNASSIGNED &&
+ icu_unicode_version < pg_unicode_version)
+ {
+ icu_skipped_codepoints++;
+ continue;
+ }
+
if (pg_category != icu_category)
{
- /*
- * A version mismatch means that some assigned codepoints in the
- * newer version may be unassigned in the older version. That's
- * OK, though the test will not cover those codepoints marked
- * unassigned in the older version (that is, it will no longer be
- * an exhaustive test).
- */
- if (pg_category == PG_U_UNASSIGNED &&
- pg_unicode_version < icu_unicode_version)
- pg_skipped_codepoints++;
- else if (icu_category == PG_U_UNASSIGNED &&
- icu_unicode_version < pg_unicode_version)
- icu_skipped_codepoints++;
- else
- {
- printf("category_test: FAILURE for codepoint 0x%06x\n", code);
- printf("category_test: Postgres category: %02d %s %s\n", pg_category,
- unicode_category_abbrev(pg_category),
- unicode_category_string(pg_category));
- printf("category_test: ICU category: %02d %s %s\n", icu_category,
- unicode_category_abbrev(icu_category),
- unicode_category_string(icu_category));
- printf("\n");
- exit(1);
- }
+ printf("category_test: FAILURE for codepoint 0x%06x\n", code);
+ printf("category_test: Postgres category: %02d %s %s\n", pg_category,
+ unicode_category_abbrev(pg_category),
+ unicode_category_string(pg_category));
+ printf("category_test: ICU category: %02d %s %s\n", icu_category,
+ unicode_category_abbrev(icu_category),
+ unicode_category_string(icu_category));
+ printf("\n");
+ exit(1);
+ }
+
+ if (prop_alphabetic != icu_prop_alphabetic ||
+ prop_lowercase != icu_prop_lowercase ||
+ prop_uppercase != icu_prop_uppercase ||
+ prop_white_space != icu_prop_white_space ||
+ prop_hex_digit != icu_prop_hex_digit ||
+ prop_join_control != icu_prop_join_control)
+ {
+ printf("category_test: FAILURE for codepoint 0x%06x\n", code);
+ printf("category_test: Postgres property alphabetic/lowercase/uppercase/white_space/hex_digit/join_control: %d/%d/%d/%d/%d/%d\n",
+ prop_alphabetic, prop_lowercase, prop_uppercase,
+ prop_white_space, prop_hex_digit, prop_join_control);
+ printf("category_test: ICU property alphabetic/lowercase/uppercase/white_space/hex_digit/join_control: %d/%d/%d/%d/%d/%d\n",
+ icu_prop_alphabetic, icu_prop_lowercase, icu_prop_uppercase,
+ icu_prop_white_space, icu_prop_hex_digit, icu_prop_join_control);
+ printf("\n");
+ exit(1);
+ }
+
+ if (isalpha != icu_isalpha ||
+ islower != icu_islower ||
+ isupper != icu_isupper ||
+ ispunct != icu_ispunct ||
+ isdigit != icu_isdigit ||
+ isxdigit != icu_isxdigit ||
+ isalnum != icu_isalnum ||
+ isspace != icu_isspace ||
+ isblank != icu_isblank ||
+ iscntrl != icu_iscntrl ||
+ isgraph != icu_isgraph ||
+ isprint != icu_isprint)
+ {
+ printf("category_test: FAILURE for codepoint 0x%06x\n", code);
+ printf("category_test: Postgres class alpha/lower/upper/punct/digit/xdigit/alnum/space/blank/cntrl/graph/print: %d/%d/%d/%d/%d/%d/%d/%d/%d/%d/%d/%d\n",
+ isalpha, islower, isupper, ispunct, isdigit, isxdigit, isalnum, isspace, isblank, iscntrl, isgraph, isprint);
+ printf("category_test: ICU class alpha/lower/upper/punct/digit/xdigit/alnum/space/blank/cntrl/graph/print: %d/%d/%d/%d/%d/%d/%d/%d/%d/%d/%d/%d\n",
+ icu_isalpha, icu_islower, icu_isupper, icu_ispunct, icu_isdigit, icu_isxdigit, icu_isalnum, icu_isspace, icu_isblank, icu_iscntrl, icu_isgraph, icu_isprint);
+ printf("\n");
+ exit(1);
}
}
@@ -99,10 +211,104 @@ main(int argc, char **argv)
printf("category_test: skipped %d codepoints unassigned in ICU due to Unicode version mismatch\n",
icu_skipped_codepoints);
- printf("category_test: success\n");
- exit(0);
+ printf("category_test: ICU test successful\n");
+}
+#endif
+
+/*
+ * For libc, test only some characters for membership in the punctuation
+ * class. We have no guarantee that all characters will obey the same rules as
+ * pg_u_ispunct_posix(), though some coverage is still useful.
+ */
+static const unsigned char test_punct[] = {
+ ',', '$', '"', 0x85, 0x00, 'b', '&', 'Z', ' ', '\t', '\n'
+};
+
+/*
+ * Test what we can for libc, which is limited but still useful to cover the
+ * _posix-variant functions.
+ */
+static void
+libc_test()
+{
+ char * libc_locale = setlocale(LC_CTYPE, LIBC_LOCALE);
+
+ if (!libc_locale)
+ {
+ printf("category_test: libc locale \"%s\" not available; skipping\n", LIBC_LOCALE);
+ return;
+ }
+
+ /* non-exhaustive test of pg_u_ispunct_posix() */
+ for (int i = 0; i < sizeof(test_punct)/sizeof(test_punct[0]); i++)
+ {
+ pg_wchar code = (pg_wchar) test_punct[i];
+ bool ispunct = pg_u_ispunct_posix(code);
+ bool libc_ispunct = iswpunct(code);
+
+ if (ispunct != libc_ispunct)
+ {
+ printf("category_test: FAILURE for codepoint 0x%06x\n", code);
+ printf("category_test: Postgres ispunct_posix: %d\n", ispunct);
+ printf("category_test: libc iswpunct: %d\n", libc_ispunct);
+ printf("\n");
+ exit(1);
+ }
+ }
+
+ for (pg_wchar code = 0; code <= 0x10ffff; code++)
+ {
+ uint8_t pg_category = unicode_category(code);
+
+ bool isalpha = pg_u_isalpha(code);
+ bool isdigit = pg_u_isdigit_posix(code);
+ bool isxdigit = pg_u_isxdigit_posix(code);
+ bool isalnum = pg_u_isalnum_posix(code);
+
+ bool libc_isdigit = iswdigit(code);
+ bool libc_isxdigit = iswxdigit(code);
+
+ if (pg_category == PG_U_UNASSIGNED)
+ continue;
+
+ /* check that alnum is the same as isdigit OR isalpha */
+ if (((isdigit || isalpha) && !isalnum) ||
+ (!(isdigit || isalpha) && isalnum))
+ {
+ printf("category_test: FAILURE for codepoint 0x%06x\n", code);
+ printf("category_test: isalnum inconsistent: isalpha/isdigit/isalnum: %d/%d/%d\n",
+ isalpha, isdigit, isalnum);
+ exit(1);
+ }
+
+ if (isdigit != libc_isdigit ||
+ isxdigit != libc_isxdigit)
+ {
+ printf("category_test: FAILURE for codepoint 0x%06x\n", code);
+ printf("category_test: Postgres class digit/xdigit: %d/%d\n",
+ isdigit, isxdigit);
+ printf("category_test: libc class digit/xdigit: %d/%d\n",
+ libc_isdigit, libc_isxdigit);
+ printf("\n");
+ exit(1);
+ }
+ }
+}
+
+int
+main(int argc, char **argv)
+{
+ pg_unicode_version = parse_unicode_version(PG_UNICODE_VERSION);
+ printf("category_test: Postgres Unicode version:\t%s\n", PG_UNICODE_VERSION);
+
+ libc_test();
+
+#ifdef USE_ICU
+ icu_unicode_version = parse_unicode_version(U_UNICODE_VERSION);
+ printf("category_test: ICU Unicode version:\t\t%s\n", U_UNICODE_VERSION);
+
+ icu_test();
#else
- printf("category_test: ICU support required for test; skipping\n");
- exit(0);
+ printf("category_test: ICU not available; skipping\n");
#endif
}
diff --git a/src/common/unicode/generate-unicode_category_table.pl b/src/common/unicode/generate-unicode_category_table.pl
index 992b877ede..9545728443 100644
--- a/src/common/unicode/generate-unicode_category_table.pl
+++ b/src/common/unicode/generate-unicode_category_table.pl
@@ -120,8 +120,6 @@ if ($range_category ne $CATEGORY_UNASSIGNED) {
category => $range_category});
}
-my $num_ranges = scalar @category_ranges;
-
# See: https://www.unicode.org/reports/tr44/#General_Category_Values
my $categories = {
Cn => 'PG_U_UNASSIGNED',
@@ -156,11 +154,98 @@ my $categories = {
Pf => 'PG_U_FINAL_PUNCTUATION'
};
-# Start writing out the output files
+# Find White_Space and Hex_Digit characters
+my @white_space = ();
+my @hex_digits = ();
+my @join_control = ();
+open($FH, '<', "$output_path/PropList.txt")
+ or die "Could not open $output_path/PropList.txt: $!.";
+while (my $line = <$FH>)
+{
+ my $pattern = qr/([0-9A-F\.]+)\s*;\s*(\w+)\s*#.*/s;
+ next unless $line =~ $pattern;
+
+ my $code = $line =~ s/$pattern/$1/rg;
+ my $property = $line =~ s/$pattern/$2/rg;
+ my $start;
+ my $end;
+
+ if ($code =~ /\.\./) {
+ # code range
+ my @sp = split /\.\./, $code;
+ $start = hex($sp[0]);
+ $end = hex($sp[1]);
+ } else {
+ # single code point
+ $start = hex($code);
+ $end = hex($code);
+ }
+
+ if ($property eq "White_Space") {
+ push @white_space, {start => $start, end => $end};
+ }
+ elsif ($property eq "Hex_Digit") {
+ push @hex_digits, {start => $start, end => $end};
+ }
+ elsif ($property eq "Join_Control") {
+ push @join_control, {start => $start, end => $end};
+ }
+}
+
+# Find Alphabetic, Lowercase, and Uppercase characters
+my @alphabetic = ();
+my @lowercase = ();
+my @uppercase = ();
+open($FH, '<', "$output_path/DerivedCoreProperties.txt")
+ or die "Could not open $output_path/DerivedCoreProperties.txt: $!.";
+while (my $line = <$FH>)
+{
+ my $pattern = qr/^([0-9A-F\.]+)\s*;\s*(\w+)\s*#.*$/s;
+ next unless $line =~ $pattern;
+
+ my $code = $line =~ s/$pattern/$1/rg;
+ my $property = $line =~ s/$pattern/$2/rg;
+ my $start;
+ my $end;
+
+ if ($code =~ /\.\./) {
+ # code range
+ my @sp = split /\.\./, $code;
+ die "line: {$line} code: {$code} sp[0] {$sp[0]} sp[1] {$sp[1]}"
+ unless $sp[0] =~ /^[0-9A-F]+$/ && $sp[1] =~ /^[0-9A-F]+$/;
+ $start = hex($sp[0]);
+ $end = hex($sp[1]);
+ } else {
+ die "line: {$line} code: {$code}" unless $code =~ /^[0-9A-F]+$/;
+ # single code point
+ $start = hex($code);
+ $end = hex($code);
+ }
+
+ if ($property eq "Alphabetic") {
+ push @alphabetic, {start => $start, end => $end};
+ }
+ elsif ($property eq "Lowercase") {
+ push @lowercase, {start => $start, end => $end};
+ }
+ elsif ($property eq "Uppercase") {
+ push @uppercase, {start => $start, end => $end};
+ }
+}
+
+my $num_category_ranges = scalar @category_ranges;
+my $num_alphabetic_ranges = scalar @alphabetic;
+my $num_lowercase_ranges = scalar @lowercase;
+my $num_uppercase_ranges = scalar @uppercase;
+my $num_white_space_ranges = scalar @white_space;
+my $num_hex_digit_ranges = scalar @hex_digits;
+my $num_join_control_ranges = scalar @join_control;
+
+# Start writing out the output file
open my $OT, '>', $output_table_file
or die "Could not open output file $output_table_file: $!\n";
-print $OT <<HEADER;
+print $OT <<"HEADER";
/*-------------------------------------------------------------------------
*
* unicode_category_table.h
@@ -188,11 +273,20 @@ typedef struct
uint8 category; /* General Category */
} pg_category_range;
-/* table of Unicode codepoint ranges and their categories */
-static const pg_category_range unicode_categories[$num_ranges] =
+typedef struct
{
+ uint32 first; /* Unicode codepoint */
+ uint32 last; /* Unicode codepoint */
+} pg_unicode_range;
+
HEADER
+print $OT <<"CATEGORY_TABLE";
+/* table of Unicode codepoint ranges and their categories */
+static const pg_category_range unicode_categories[$num_category_ranges] =
+{
+CATEGORY_TABLE
+
my $firsttime = 1;
foreach my $range (@category_ranges) {
printf $OT ",\n" unless $firsttime;
@@ -202,4 +296,101 @@ foreach my $range (@category_ranges) {
die "category missing: $range->{category}" unless $category;
printf $OT "\t{0x%06x, 0x%06x, %s}", $range->{start}, $range->{end}, $category;
}
+
+print $OT "\n};\n\n";
+
+print $OT <<"ALPHABETIC_TABLE";
+/* table of Unicode codepoint ranges of Alphabetic characters */
+static const pg_unicode_range unicode_alphabetic[$num_alphabetic_ranges] =
+{
+ALPHABETIC_TABLE
+
+$firsttime = 1;
+foreach my $range (@alphabetic) {
+ printf $OT ",\n" unless $firsttime;
+ $firsttime = 0;
+
+ printf $OT "\t{0x%06x, 0x%06x}", $range->{start}, $range->{end};
+}
+
+print $OT "\n};\n\n";
+
+print $OT <<"LOWERCASE_TABLE";
+/* table of Unicode codepoint ranges of Lowercase characters */
+static const pg_unicode_range unicode_lowercase[$num_lowercase_ranges] =
+{
+LOWERCASE_TABLE
+
+$firsttime = 1;
+foreach my $range (@lowercase) {
+ printf $OT ",\n" unless $firsttime;
+ $firsttime = 0;
+
+ printf $OT "\t{0x%06x, 0x%06x}", $range->{start}, $range->{end};
+}
+
+print $OT "\n};\n\n";
+
+print $OT <<"UPPERCASE_TABLE";
+/* table of Unicode codepoint ranges of Uppercase characters */
+static const pg_unicode_range unicode_uppercase[$num_uppercase_ranges] =
+{
+UPPERCASE_TABLE
+
+$firsttime = 1;
+foreach my $range (@uppercase) {
+ printf $OT ",\n" unless $firsttime;
+ $firsttime = 0;
+
+ printf $OT "\t{0x%06x, 0x%06x}", $range->{start}, $range->{end};
+}
+
+print $OT "\n};\n\n";
+
+print $OT <<"WHITE_SPACE_TABLE";
+/* table of Unicode codepoint ranges of White_Space characters */
+static const pg_unicode_range unicode_white_space[$num_white_space_ranges] =
+{
+WHITE_SPACE_TABLE
+
+$firsttime = 1;
+foreach my $range (@white_space) {
+ printf $OT ",\n" unless $firsttime;
+ $firsttime = 0;
+
+ printf $OT "\t{0x%06x, 0x%06x}", $range->{start}, $range->{end};
+}
+
+print $OT "\n};\n\n";
+
+print $OT <<"HEX_DIGITS_TABLE";
+/* table of Unicode codepoint ranges of Hex_Digit characters */
+static const pg_unicode_range unicode_hex_digit[$num_hex_digit_ranges] =
+{
+HEX_DIGITS_TABLE
+
+$firsttime = 1;
+foreach my $range (@hex_digits) {
+ printf $OT ",\n" unless $firsttime;
+ $firsttime = 0;
+
+ printf $OT "\t{0x%06x, 0x%06x}", $range->{start}, $range->{end};
+}
+
+print $OT "\n};\n\n";
+
+print $OT <<"JOIN_CONTROL_TABLE";
+/* table of Unicode codepoint ranges of Join_Control characters */
+static const pg_unicode_range unicode_join_control[$num_join_control_ranges] =
+{
+JOIN_CONTROL_TABLE
+
+$firsttime = 1;
+foreach my $range (@join_control) {
+ printf $OT ",\n" unless $firsttime;
+ $firsttime = 0;
+
+ printf $OT "\t{0x%06x, 0x%06x}", $range->{start}, $range->{end};
+}
+
print $OT "\n};\n";
diff --git a/src/common/unicode/meson.build b/src/common/unicode/meson.build
index e8cfdc1df4..3526ddb846 100644
--- a/src/common/unicode/meson.build
+++ b/src/common/unicode/meson.build
@@ -11,7 +11,7 @@ endif
# These files are part of the Unicode Character Database. Download them on
# demand.
-foreach f : ['CompositionExclusions.txt', 'DerivedNormalizationProps.txt', 'EastAsianWidth.txt', 'NormalizationTest.txt', 'UnicodeData.txt']
+foreach f : ['CompositionExclusions.txt', 'DerivedCoreProperties.txt', 'DerivedNormalizationProps.txt', 'EastAsianWidth.txt', 'NormalizationTest.txt', 'PropList.txt', 'UnicodeData.txt']
url = unicode_baseurl.format(UNICODE_VERSION, f)
target = custom_target(f,
output: f,
@@ -26,7 +26,7 @@ update_unicode_targets = []
update_unicode_targets += \
custom_target('unicode_category_table.h',
- input: [unicode_data['UnicodeData.txt']],
+ input: [unicode_data['UnicodeData.txt'], unicode_data['DerivedCoreProperties.txt'], unicode_data['PropList.txt']],
output: ['unicode_category_table.h'],
command: [
perl, files('generate-unicode_category_table.pl'),
diff --git a/src/common/unicode_category.c b/src/common/unicode_category.c
index 189cd6eca3..efe617d45b 100644
--- a/src/common/unicode_category.c
+++ b/src/common/unicode_category.c
@@ -1,6 +1,8 @@
/*-------------------------------------------------------------------------
* unicode_category.c
- * Determine general category of Unicode characters.
+ * Determine general category and character class of Unicode
+ * characters. Encoding must be UTF8, where we assume that the pg_wchar
+ * representation is a code point.
*
* Portions Copyright (c) 2017-2023, PostgreSQL Global Development Group
*
@@ -18,24 +20,78 @@
#include "common/unicode_category.h"
#include "common/unicode_category_table.h"
+/*
+ * We use a mask word for convenience when testing for multiple categories at
+ * once. The number of Unicode General Categories should never grow, so a
+ * 32-bit mask is fine.
+ */
+#define PG_U_CATEGORY_MASK(X) ((uint32)(1 << (X)))
+
+#define PG_U_LU_MASK PG_U_CATEGORY_MASK(PG_U_UPPERCASE_LETTER)
+#define PG_U_LL_MASK PG_U_CATEGORY_MASK(PG_U_LOWERCASE_LETTER)
+#define PG_U_LT_MASK PG_U_CATEGORY_MASK(PG_U_TITLECASE_LETTER)
+#define PG_U_LC_MASK (PG_U_LU_MASK|PG_U_LL_MASK|PG_U_LT_MASK)
+#define PG_U_LM_MASK PG_U_CATEGORY_MASK(PG_U_MODIFIER_LETTER)
+#define PG_U_LO_MASK PG_U_CATEGORY_MASK(PG_U_OTHER_LETTER)
+#define PG_U_L_MASK (PG_U_LU_MASK|PG_U_LL_MASK|PG_U_LT_MASK|PG_U_LM_MASK|\
+ PG_U_LO_MASK)
+#define PG_U_MN_MASK PG_U_CATEGORY_MASK(PG_U_NONSPACING_MARK)
+#define PG_U_ME_MASK PG_U_CATEGORY_MASK(PG_U_ENCLOSING_MARK)
+#define PG_U_MC_MASK PG_U_CATEGORY_MASK(PG_U_SPACING_MARK)
+#define PG_U_M_MASK (PG_U_MN_MASK|PG_U_MC_MASK|PG_U_ME_MASK)
+#define PG_U_ND_MASK PG_U_CATEGORY_MASK(PG_U_DECIMAL_NUMBER)
+#define PG_U_NL_MASK PG_U_CATEGORY_MASK(PG_U_LETTER_NUMBER)
+#define PG_U_NO_MASK PG_U_CATEGORY_MASK(PG_U_OTHER_NUMBER)
+#define PG_U_N_MASK (PG_U_ND_MASK|PG_U_NL_MASK|PG_U_NO_MASK)
+#define PG_U_PC_MASK PG_U_CATEGORY_MASK(PG_U_CONNECTOR_PUNCTUATION)
+#define PG_U_PD_MASK PG_U_CATEGORY_MASK(PG_U_DASH_PUNCTUATION)
+#define PG_U_PS_MASK PG_U_CATEGORY_MASK(PG_U_OPEN_PUNCTUATION)
+#define PG_U_PE_MASK PG_U_CATEGORY_MASK(PG_U_CLOSE_PUNCTUATION)
+#define PG_U_PI_MASK PG_U_CATEGORY_MASK(PG_U_INITIAL_PUNCTUATION)
+#define PG_U_PF_MASK PG_U_CATEGORY_MASK(PG_U_FINAL_PUNCTUATION)
+#define PG_U_PO_MASK PG_U_CATEGORY_MASK(PG_U_OTHER_PUNCTUATION)
+#define PG_U_P_MASK (PG_U_PC_MASK|PG_U_PD_MASK|PG_U_PS_MASK|PG_U_PE_MASK|\
+ PG_U_PI_MASK|PG_U_PF_MASK|PG_U_PO_MASK)
+#define PG_U_SM_MASK PG_U_CATEGORY_MASK(PG_U_MATH_SYMBOL)
+#define PG_U_SC_MASK PG_U_CATEGORY_MASK(PG_U_CURRENCY_SYMBOL)
+#define PG_U_SK_MASK PG_U_CATEGORY_MASK(PG_U_MODIFIER_SYMBOL)
+#define PG_U_SO_MASK PG_U_CATEGORY_MASK(PG_U_OTHER_SYMBOL)
+#define PG_U_S_MASK (PG_U_SM_MASK|PG_U_SC_MASK|PG_U_SK_MASK|PG_U_SO_MASK)
+#define PG_U_ZS_MASK PG_U_CATEGORY_MASK(PG_U_SPACE_SEPARATOR)
+#define PG_U_ZL_MASK PG_U_CATEGORY_MASK(PG_U_LINE_SEPARATOR)
+#define PG_U_ZP_MASK PG_U_CATEGORY_MASK(PG_U_PARAGRAPH_SEPARATOR)
+#define PG_U_Z_MASK (PG_U_ZS_MASK|PG_U_ZL_MASK|PG_U_ZP_MASK)
+#define PG_U_CC_MASK PG_U_CATEGORY_MASK(PG_U_CONTROL)
+#define PG_U_CF_MASK PG_U_CATEGORY_MASK(PG_U_FORMAT)
+#define PG_U_CS_MASK PG_U_CATEGORY_MASK(PG_U_SURROGATE)
+#define PG_U_CO_MASK PG_U_CATEGORY_MASK(PG_U_PRIVATE_USE)
+#define PG_U_CN_MASK PG_U_CATEGORY_MASK(PG_U_UNASSIGNED)
+#define PG_U_C_MASK (PG_U_CC_MASK|PG_U_CF_MASK|PG_U_CS_MASK|PG_U_CO_MASK|\
+ PG_U_CN_MASK)
+
+#define PG_U_CHARACTER_TAB 0x09
+
+static bool range_search(const pg_unicode_range * tbl, Size size,
+ pg_wchar code);
+
/*
* Unicode general category for the given codepoint.
*/
pg_unicode_category
-unicode_category(pg_wchar ucs)
+unicode_category(pg_wchar code)
{
int min = 0;
int mid;
int max = lengthof(unicode_categories) - 1;
- Assert(ucs <= 0x10ffff);
+ Assert(code <= 0x10ffff);
while (max >= min)
{
mid = (min + max) / 2;
- if (ucs > unicode_categories[mid].last)
+ if (code > unicode_categories[mid].last)
min = mid + 1;
- else if (ucs < unicode_categories[mid].first)
+ else if (code < unicode_categories[mid].first)
max = mid - 1;
else
return unicode_categories[mid].category;
@@ -44,6 +100,179 @@ unicode_category(pg_wchar ucs)
return PG_U_UNASSIGNED;
}
+bool
+pg_u_prop_alphabetic(pg_wchar code)
+{
+ return range_search(unicode_alphabetic, lengthof(unicode_alphabetic),
+ code);
+}
+
+bool
+pg_u_prop_lowercase(pg_wchar code)
+{
+ return range_search(unicode_lowercase, lengthof(unicode_lowercase), code);
+}
+
+bool
+pg_u_prop_uppercase(pg_wchar code)
+{
+ return range_search(unicode_uppercase, lengthof(unicode_uppercase), code);
+}
+
+bool
+pg_u_prop_white_space(pg_wchar code)
+{
+ return range_search(unicode_white_space, lengthof(unicode_white_space),
+ code);
+}
+
+bool
+pg_u_prop_hex_digit(pg_wchar code)
+{
+ return range_search(unicode_hex_digit, lengthof(unicode_hex_digit), code);
+}
+
+bool
+pg_u_prop_join_control(pg_wchar code)
+{
+ return range_search(unicode_join_control, lengthof(unicode_join_control),
+ code);
+}
+
+/*
+ * The following functions implement the regex character classification as
+ * described at: http://www.unicode.org/reports/tr18/#Compatibility_Properties
+ */
+
+bool
+pg_u_isdigit(pg_wchar code)
+{
+ return unicode_category(code) == PG_U_DECIMAL_NUMBER;
+}
+
+bool
+pg_u_isdigit_posix(pg_wchar code)
+{
+ return ('0' <= code && code <= '9');
+}
+
+bool
+pg_u_isalpha(pg_wchar code)
+{
+ return pg_u_prop_alphabetic(code);
+}
+
+bool
+pg_u_isalnum(pg_wchar code)
+{
+ return pg_u_isalpha(code) || pg_u_isdigit(code);
+}
+
+bool
+pg_u_isalnum_posix(pg_wchar code)
+{
+ return pg_u_isalpha(code) || pg_u_isdigit_posix(code);
+}
+
+bool
+pg_u_isword(pg_wchar code)
+{
+ uint32 category_mask = PG_U_CATEGORY_MASK(unicode_category(code));
+
+ return
+ category_mask & (PG_U_M_MASK|PG_U_ND_MASK|PG_U_PC_MASK) ||
+ pg_u_isalpha(code) ||
+ pg_u_prop_join_control(code);
+}
+
+bool
+pg_u_isupper(pg_wchar code)
+{
+ return pg_u_prop_uppercase(code);
+}
+
+bool
+pg_u_islower(pg_wchar code)
+{
+ return pg_u_prop_lowercase(code);
+}
+
+bool
+pg_u_isblank(pg_wchar code)
+{
+ return code == PG_U_CHARACTER_TAB ||
+ unicode_category(code) == PG_U_SPACE_SEPARATOR;
+}
+
+bool
+pg_u_iscntrl(pg_wchar code)
+{
+ return unicode_category(code) == PG_U_CONTROL;
+}
+
+bool
+pg_u_isgraph(pg_wchar code)
+{
+ uint32 category_mask = PG_U_CATEGORY_MASK(unicode_category(code));
+
+ if (category_mask & (PG_U_CC_MASK|PG_U_CS_MASK|PG_U_CN_MASK) ||
+ pg_u_isspace(code))
+ return false;
+ return true;
+}
+
+bool
+pg_u_isprint(pg_wchar code)
+{
+ pg_unicode_category category = unicode_category(code);
+
+ if (category == PG_U_CONTROL)
+ return false;
+
+ return pg_u_isgraph(code) || pg_u_isblank(code);
+}
+
+bool
+pg_u_ispunct(pg_wchar code)
+{
+ uint32 category_mask = PG_U_CATEGORY_MASK(unicode_category(code));
+
+ return category_mask & PG_U_P_MASK;
+}
+
+bool
+pg_u_ispunct_posix(pg_wchar code)
+{
+ uint32 category_mask;
+
+ if (pg_u_isalpha(code))
+ return false;
+
+ category_mask = PG_U_CATEGORY_MASK(unicode_category(code));
+ return category_mask & (PG_U_P_MASK|PG_U_S_MASK);
+}
+
+bool
+pg_u_isspace(pg_wchar code)
+{
+ return pg_u_prop_white_space(code);
+}
+
+bool
+pg_u_isxdigit(pg_wchar code)
+{
+ return unicode_category(code) == PG_U_DECIMAL_NUMBER ||
+ pg_u_prop_hex_digit(code);
+}
+
+bool
+pg_u_isxdigit_posix(pg_wchar code)
+{
+ return (('0' <= code && code <= '9') ||
+ ('A' <= code && code <= 'F') ||
+ ('a' <= code && code <= 'f'));
+}
+
/*
* Description of Unicode general category.
*/
@@ -191,3 +420,30 @@ unicode_category_abbrev(pg_unicode_category category)
Assert(false);
return "??"; /* keep compiler quiet */
}
+
+/*
+ * Binary search to test if given codepoint exists in one of the ranges in the
+ * given table.
+ */
+static bool
+range_search(const pg_unicode_range * tbl, Size size, pg_wchar code)
+{
+ int min = 0;
+ int mid;
+ int max = size - 1;
+
+ Assert(code <= 0x10ffff);
+
+ while (max >= min)
+ {
+ mid = (min + max) / 2;
+ if (code > tbl[mid].last)
+ min = mid + 1;
+ else if (code < tbl[mid].first)
+ max = mid - 1;
+ else
+ return true;
+ }
+
+ return false;
+}
diff --git a/src/include/common/unicode_category.h b/src/include/common/unicode_category.h
index 81d38c7411..7d9ef2b496 100644
--- a/src/include/common/unicode_category.h
+++ b/src/include/common/unicode_category.h
@@ -62,7 +62,32 @@ typedef enum pg_unicode_category
} pg_unicode_category;
extern pg_unicode_category unicode_category(pg_wchar ucs);
-const char *unicode_category_string(pg_unicode_category category);
-const char *unicode_category_abbrev(pg_unicode_category category);
+extern const char *unicode_category_string(pg_unicode_category category);
+extern const char *unicode_category_abbrev(pg_unicode_category category);
+
+extern bool pg_u_prop_alphabetic(pg_wchar c);
+extern bool pg_u_prop_lowercase(pg_wchar c);
+extern bool pg_u_prop_uppercase(pg_wchar c);
+extern bool pg_u_prop_white_space(pg_wchar c);
+extern bool pg_u_prop_hex_digit(pg_wchar c);
+extern bool pg_u_prop_join_control(pg_wchar c);
+
+extern bool pg_u_isdigit(pg_wchar c);
+extern bool pg_u_isdigit_posix(pg_wchar c);
+extern bool pg_u_isalpha(pg_wchar c);
+extern bool pg_u_isalnum(pg_wchar c);
+extern bool pg_u_isalnum_posix(pg_wchar c);
+extern bool pg_u_isword(pg_wchar c);
+extern bool pg_u_isupper(pg_wchar c);
+extern bool pg_u_islower(pg_wchar c);
+extern bool pg_u_isblank(pg_wchar c);
+extern bool pg_u_iscntrl(pg_wchar c);
+extern bool pg_u_isgraph(pg_wchar c);
+extern bool pg_u_isprint(pg_wchar c);
+extern bool pg_u_ispunct(pg_wchar c);
+extern bool pg_u_ispunct_posix(pg_wchar c);
+extern bool pg_u_isspace(pg_wchar c);
+extern bool pg_u_isxdigit(pg_wchar c);
+extern bool pg_u_isxdigit_posix(pg_wchar c);
#endif /* UNICODE_CATEGORY_H */
diff --git a/src/include/common/unicode_category_table.h b/src/include/common/unicode_category_table.h
index 14f1ea0677..86cdc9c0ed 100644
--- a/src/include/common/unicode_category_table.h
+++ b/src/include/common/unicode_category_table.h
@@ -25,6 +25,12 @@ typedef struct
uint8 category; /* General Category */
} pg_category_range;
+typedef struct
+{
+ uint32 first; /* Unicode codepoint */
+ uint32 last; /* Unicode codepoint */
+} pg_unicode_range;
+
/* table of Unicode codepoint ranges and their categories */
static const pg_category_range unicode_categories[3302] =
{
@@ -3331,3 +3337,2529 @@ static const pg_category_range unicode_categories[3302] =
{0x0f0000, 0x0ffffd, PG_U_PRIVATE_USE},
{0x100000, 0x10fffd, PG_U_PRIVATE_USE}
};
+
+/* table of Unicode codepoint ranges of Alphabetic characters */
+static const pg_unicode_range unicode_alphabetic[1141] =
+{
+ {0x000041, 0x00005a},
+ {0x000061, 0x00007a},
+ {0x0000aa, 0x0000aa},
+ {0x0000b5, 0x0000b5},
+ {0x0000ba, 0x0000ba},
+ {0x0000c0, 0x0000d6},
+ {0x0000d8, 0x0000f6},
+ {0x0000f8, 0x0001ba},
+ {0x0001bb, 0x0001bb},
+ {0x0001bc, 0x0001bf},
+ {0x0001c0, 0x0001c3},
+ {0x0001c4, 0x000293},
+ {0x000294, 0x000294},
+ {0x000295, 0x0002af},
+ {0x0002b0, 0x0002c1},
+ {0x0002c6, 0x0002d1},
+ {0x0002e0, 0x0002e4},
+ {0x0002ec, 0x0002ec},
+ {0x0002ee, 0x0002ee},
+ {0x000345, 0x000345},
+ {0x000370, 0x000373},
+ {0x000374, 0x000374},
+ {0x000376, 0x000377},
+ {0x00037a, 0x00037a},
+ {0x00037b, 0x00037d},
+ {0x00037f, 0x00037f},
+ {0x000386, 0x000386},
+ {0x000388, 0x00038a},
+ {0x00038c, 0x00038c},
+ {0x00038e, 0x0003a1},
+ {0x0003a3, 0x0003f5},
+ {0x0003f7, 0x000481},
+ {0x00048a, 0x00052f},
+ {0x000531, 0x000556},
+ {0x000559, 0x000559},
+ {0x000560, 0x000588},
+ {0x0005b0, 0x0005bd},
+ {0x0005bf, 0x0005bf},
+ {0x0005c1, 0x0005c2},
+ {0x0005c4, 0x0005c5},
+ {0x0005c7, 0x0005c7},
+ {0x0005d0, 0x0005ea},
+ {0x0005ef, 0x0005f2},
+ {0x000610, 0x00061a},
+ {0x000620, 0x00063f},
+ {0x000640, 0x000640},
+ {0x000641, 0x00064a},
+ {0x00064b, 0x000657},
+ {0x000659, 0x00065f},
+ {0x00066e, 0x00066f},
+ {0x000670, 0x000670},
+ {0x000671, 0x0006d3},
+ {0x0006d5, 0x0006d5},
+ {0x0006d6, 0x0006dc},
+ {0x0006e1, 0x0006e4},
+ {0x0006e5, 0x0006e6},
+ {0x0006e7, 0x0006e8},
+ {0x0006ed, 0x0006ed},
+ {0x0006ee, 0x0006ef},
+ {0x0006fa, 0x0006fc},
+ {0x0006ff, 0x0006ff},
+ {0x000710, 0x000710},
+ {0x000711, 0x000711},
+ {0x000712, 0x00072f},
+ {0x000730, 0x00073f},
+ {0x00074d, 0x0007a5},
+ {0x0007a6, 0x0007b0},
+ {0x0007b1, 0x0007b1},
+ {0x0007ca, 0x0007ea},
+ {0x0007f4, 0x0007f5},
+ {0x0007fa, 0x0007fa},
+ {0x000800, 0x000815},
+ {0x000816, 0x000817},
+ {0x00081a, 0x00081a},
+ {0x00081b, 0x000823},
+ {0x000824, 0x000824},
+ {0x000825, 0x000827},
+ {0x000828, 0x000828},
+ {0x000829, 0x00082c},
+ {0x000840, 0x000858},
+ {0x000860, 0x00086a},
+ {0x000870, 0x000887},
+ {0x000889, 0x00088e},
+ {0x0008a0, 0x0008c8},
+ {0x0008c9, 0x0008c9},
+ {0x0008d4, 0x0008df},
+ {0x0008e3, 0x0008e9},
+ {0x0008f0, 0x000902},
+ {0x000903, 0x000903},
+ {0x000904, 0x000939},
+ {0x00093a, 0x00093a},
+ {0x00093b, 0x00093b},
+ {0x00093d, 0x00093d},
+ {0x00093e, 0x000940},
+ {0x000941, 0x000948},
+ {0x000949, 0x00094c},
+ {0x00094e, 0x00094f},
+ {0x000950, 0x000950},
+ {0x000955, 0x000957},
+ {0x000958, 0x000961},
+ {0x000962, 0x000963},
+ {0x000971, 0x000971},
+ {0x000972, 0x000980},
+ {0x000981, 0x000981},
+ {0x000982, 0x000983},
+ {0x000985, 0x00098c},
+ {0x00098f, 0x000990},
+ {0x000993, 0x0009a8},
+ {0x0009aa, 0x0009b0},
+ {0x0009b2, 0x0009b2},
+ {0x0009b6, 0x0009b9},
+ {0x0009bd, 0x0009bd},
+ {0x0009be, 0x0009c0},
+ {0x0009c1, 0x0009c4},
+ {0x0009c7, 0x0009c8},
+ {0x0009cb, 0x0009cc},
+ {0x0009ce, 0x0009ce},
+ {0x0009d7, 0x0009d7},
+ {0x0009dc, 0x0009dd},
+ {0x0009df, 0x0009e1},
+ {0x0009e2, 0x0009e3},
+ {0x0009f0, 0x0009f1},
+ {0x0009fc, 0x0009fc},
+ {0x000a01, 0x000a02},
+ {0x000a03, 0x000a03},
+ {0x000a05, 0x000a0a},
+ {0x000a0f, 0x000a10},
+ {0x000a13, 0x000a28},
+ {0x000a2a, 0x000a30},
+ {0x000a32, 0x000a33},
+ {0x000a35, 0x000a36},
+ {0x000a38, 0x000a39},
+ {0x000a3e, 0x000a40},
+ {0x000a41, 0x000a42},
+ {0x000a47, 0x000a48},
+ {0x000a4b, 0x000a4c},
+ {0x000a51, 0x000a51},
+ {0x000a59, 0x000a5c},
+ {0x000a5e, 0x000a5e},
+ {0x000a70, 0x000a71},
+ {0x000a72, 0x000a74},
+ {0x000a75, 0x000a75},
+ {0x000a81, 0x000a82},
+ {0x000a83, 0x000a83},
+ {0x000a85, 0x000a8d},
+ {0x000a8f, 0x000a91},
+ {0x000a93, 0x000aa8},
+ {0x000aaa, 0x000ab0},
+ {0x000ab2, 0x000ab3},
+ {0x000ab5, 0x000ab9},
+ {0x000abd, 0x000abd},
+ {0x000abe, 0x000ac0},
+ {0x000ac1, 0x000ac5},
+ {0x000ac7, 0x000ac8},
+ {0x000ac9, 0x000ac9},
+ {0x000acb, 0x000acc},
+ {0x000ad0, 0x000ad0},
+ {0x000ae0, 0x000ae1},
+ {0x000ae2, 0x000ae3},
+ {0x000af9, 0x000af9},
+ {0x000afa, 0x000afc},
+ {0x000b01, 0x000b01},
+ {0x000b02, 0x000b03},
+ {0x000b05, 0x000b0c},
+ {0x000b0f, 0x000b10},
+ {0x000b13, 0x000b28},
+ {0x000b2a, 0x000b30},
+ {0x000b32, 0x000b33},
+ {0x000b35, 0x000b39},
+ {0x000b3d, 0x000b3d},
+ {0x000b3e, 0x000b3e},
+ {0x000b3f, 0x000b3f},
+ {0x000b40, 0x000b40},
+ {0x000b41, 0x000b44},
+ {0x000b47, 0x000b48},
+ {0x000b4b, 0x000b4c},
+ {0x000b56, 0x000b56},
+ {0x000b57, 0x000b57},
+ {0x000b5c, 0x000b5d},
+ {0x000b5f, 0x000b61},
+ {0x000b62, 0x000b63},
+ {0x000b71, 0x000b71},
+ {0x000b82, 0x000b82},
+ {0x000b83, 0x000b83},
+ {0x000b85, 0x000b8a},
+ {0x000b8e, 0x000b90},
+ {0x000b92, 0x000b95},
+ {0x000b99, 0x000b9a},
+ {0x000b9c, 0x000b9c},
+ {0x000b9e, 0x000b9f},
+ {0x000ba3, 0x000ba4},
+ {0x000ba8, 0x000baa},
+ {0x000bae, 0x000bb9},
+ {0x000bbe, 0x000bbf},
+ {0x000bc0, 0x000bc0},
+ {0x000bc1, 0x000bc2},
+ {0x000bc6, 0x000bc8},
+ {0x000bca, 0x000bcc},
+ {0x000bd0, 0x000bd0},
+ {0x000bd7, 0x000bd7},
+ {0x000c00, 0x000c00},
+ {0x000c01, 0x000c03},
+ {0x000c04, 0x000c04},
+ {0x000c05, 0x000c0c},
+ {0x000c0e, 0x000c10},
+ {0x000c12, 0x000c28},
+ {0x000c2a, 0x000c39},
+ {0x000c3d, 0x000c3d},
+ {0x000c3e, 0x000c40},
+ {0x000c41, 0x000c44},
+ {0x000c46, 0x000c48},
+ {0x000c4a, 0x000c4c},
+ {0x000c55, 0x000c56},
+ {0x000c58, 0x000c5a},
+ {0x000c5d, 0x000c5d},
+ {0x000c60, 0x000c61},
+ {0x000c62, 0x000c63},
+ {0x000c80, 0x000c80},
+ {0x000c81, 0x000c81},
+ {0x000c82, 0x000c83},
+ {0x000c85, 0x000c8c},
+ {0x000c8e, 0x000c90},
+ {0x000c92, 0x000ca8},
+ {0x000caa, 0x000cb3},
+ {0x000cb5, 0x000cb9},
+ {0x000cbd, 0x000cbd},
+ {0x000cbe, 0x000cbe},
+ {0x000cbf, 0x000cbf},
+ {0x000cc0, 0x000cc4},
+ {0x000cc6, 0x000cc6},
+ {0x000cc7, 0x000cc8},
+ {0x000cca, 0x000ccb},
+ {0x000ccc, 0x000ccc},
+ {0x000cd5, 0x000cd6},
+ {0x000cdd, 0x000cde},
+ {0x000ce0, 0x000ce1},
+ {0x000ce2, 0x000ce3},
+ {0x000cf1, 0x000cf2},
+ {0x000cf3, 0x000cf3},
+ {0x000d00, 0x000d01},
+ {0x000d02, 0x000d03},
+ {0x000d04, 0x000d0c},
+ {0x000d0e, 0x000d10},
+ {0x000d12, 0x000d3a},
+ {0x000d3d, 0x000d3d},
+ {0x000d3e, 0x000d40},
+ {0x000d41, 0x000d44},
+ {0x000d46, 0x000d48},
+ {0x000d4a, 0x000d4c},
+ {0x000d4e, 0x000d4e},
+ {0x000d54, 0x000d56},
+ {0x000d57, 0x000d57},
+ {0x000d5f, 0x000d61},
+ {0x000d62, 0x000d63},
+ {0x000d7a, 0x000d7f},
+ {0x000d81, 0x000d81},
+ {0x000d82, 0x000d83},
+ {0x000d85, 0x000d96},
+ {0x000d9a, 0x000db1},
+ {0x000db3, 0x000dbb},
+ {0x000dbd, 0x000dbd},
+ {0x000dc0, 0x000dc6},
+ {0x000dcf, 0x000dd1},
+ {0x000dd2, 0x000dd4},
+ {0x000dd6, 0x000dd6},
+ {0x000dd8, 0x000ddf},
+ {0x000df2, 0x000df3},
+ {0x000e01, 0x000e30},
+ {0x000e31, 0x000e31},
+ {0x000e32, 0x000e33},
+ {0x000e34, 0x000e3a},
+ {0x000e40, 0x000e45},
+ {0x000e46, 0x000e46},
+ {0x000e4d, 0x000e4d},
+ {0x000e81, 0x000e82},
+ {0x000e84, 0x000e84},
+ {0x000e86, 0x000e8a},
+ {0x000e8c, 0x000ea3},
+ {0x000ea5, 0x000ea5},
+ {0x000ea7, 0x000eb0},
+ {0x000eb1, 0x000eb1},
+ {0x000eb2, 0x000eb3},
+ {0x000eb4, 0x000eb9},
+ {0x000ebb, 0x000ebc},
+ {0x000ebd, 0x000ebd},
+ {0x000ec0, 0x000ec4},
+ {0x000ec6, 0x000ec6},
+ {0x000ecd, 0x000ecd},
+ {0x000edc, 0x000edf},
+ {0x000f00, 0x000f00},
+ {0x000f40, 0x000f47},
+ {0x000f49, 0x000f6c},
+ {0x000f71, 0x000f7e},
+ {0x000f7f, 0x000f7f},
+ {0x000f80, 0x000f83},
+ {0x000f88, 0x000f8c},
+ {0x000f8d, 0x000f97},
+ {0x000f99, 0x000fbc},
+ {0x001000, 0x00102a},
+ {0x00102b, 0x00102c},
+ {0x00102d, 0x001030},
+ {0x001031, 0x001031},
+ {0x001032, 0x001036},
+ {0x001038, 0x001038},
+ {0x00103b, 0x00103c},
+ {0x00103d, 0x00103e},
+ {0x00103f, 0x00103f},
+ {0x001050, 0x001055},
+ {0x001056, 0x001057},
+ {0x001058, 0x001059},
+ {0x00105a, 0x00105d},
+ {0x00105e, 0x001060},
+ {0x001061, 0x001061},
+ {0x001062, 0x001064},
+ {0x001065, 0x001066},
+ {0x001067, 0x00106d},
+ {0x00106e, 0x001070},
+ {0x001071, 0x001074},
+ {0x001075, 0x001081},
+ {0x001082, 0x001082},
+ {0x001083, 0x001084},
+ {0x001085, 0x001086},
+ {0x001087, 0x00108c},
+ {0x00108d, 0x00108d},
+ {0x00108e, 0x00108e},
+ {0x00108f, 0x00108f},
+ {0x00109a, 0x00109c},
+ {0x00109d, 0x00109d},
+ {0x0010a0, 0x0010c5},
+ {0x0010c7, 0x0010c7},
+ {0x0010cd, 0x0010cd},
+ {0x0010d0, 0x0010fa},
+ {0x0010fc, 0x0010fc},
+ {0x0010fd, 0x0010ff},
+ {0x001100, 0x001248},
+ {0x00124a, 0x00124d},
+ {0x001250, 0x001256},
+ {0x001258, 0x001258},
+ {0x00125a, 0x00125d},
+ {0x001260, 0x001288},
+ {0x00128a, 0x00128d},
+ {0x001290, 0x0012b0},
+ {0x0012b2, 0x0012b5},
+ {0x0012b8, 0x0012be},
+ {0x0012c0, 0x0012c0},
+ {0x0012c2, 0x0012c5},
+ {0x0012c8, 0x0012d6},
+ {0x0012d8, 0x001310},
+ {0x001312, 0x001315},
+ {0x001318, 0x00135a},
+ {0x001380, 0x00138f},
+ {0x0013a0, 0x0013f5},
+ {0x0013f8, 0x0013fd},
+ {0x001401, 0x00166c},
+ {0x00166f, 0x00167f},
+ {0x001681, 0x00169a},
+ {0x0016a0, 0x0016ea},
+ {0x0016ee, 0x0016f0},
+ {0x0016f1, 0x0016f8},
+ {0x001700, 0x001711},
+ {0x001712, 0x001713},
+ {0x00171f, 0x001731},
+ {0x001732, 0x001733},
+ {0x001740, 0x001751},
+ {0x001752, 0x001753},
+ {0x001760, 0x00176c},
+ {0x00176e, 0x001770},
+ {0x001772, 0x001773},
+ {0x001780, 0x0017b3},
+ {0x0017b6, 0x0017b6},
+ {0x0017b7, 0x0017bd},
+ {0x0017be, 0x0017c5},
+ {0x0017c6, 0x0017c6},
+ {0x0017c7, 0x0017c8},
+ {0x0017d7, 0x0017d7},
+ {0x0017dc, 0x0017dc},
+ {0x001820, 0x001842},
+ {0x001843, 0x001843},
+ {0x001844, 0x001878},
+ {0x001880, 0x001884},
+ {0x001885, 0x001886},
+ {0x001887, 0x0018a8},
+ {0x0018a9, 0x0018a9},
+ {0x0018aa, 0x0018aa},
+ {0x0018b0, 0x0018f5},
+ {0x001900, 0x00191e},
+ {0x001920, 0x001922},
+ {0x001923, 0x001926},
+ {0x001927, 0x001928},
+ {0x001929, 0x00192b},
+ {0x001930, 0x001931},
+ {0x001932, 0x001932},
+ {0x001933, 0x001938},
+ {0x001950, 0x00196d},
+ {0x001970, 0x001974},
+ {0x001980, 0x0019ab},
+ {0x0019b0, 0x0019c9},
+ {0x001a00, 0x001a16},
+ {0x001a17, 0x001a18},
+ {0x001a19, 0x001a1a},
+ {0x001a1b, 0x001a1b},
+ {0x001a20, 0x001a54},
+ {0x001a55, 0x001a55},
+ {0x001a56, 0x001a56},
+ {0x001a57, 0x001a57},
+ {0x001a58, 0x001a5e},
+ {0x001a61, 0x001a61},
+ {0x001a62, 0x001a62},
+ {0x001a63, 0x001a64},
+ {0x001a65, 0x001a6c},
+ {0x001a6d, 0x001a72},
+ {0x001a73, 0x001a74},
+ {0x001aa7, 0x001aa7},
+ {0x001abf, 0x001ac0},
+ {0x001acc, 0x001ace},
+ {0x001b00, 0x001b03},
+ {0x001b04, 0x001b04},
+ {0x001b05, 0x001b33},
+ {0x001b35, 0x001b35},
+ {0x001b36, 0x001b3a},
+ {0x001b3b, 0x001b3b},
+ {0x001b3c, 0x001b3c},
+ {0x001b3d, 0x001b41},
+ {0x001b42, 0x001b42},
+ {0x001b43, 0x001b43},
+ {0x001b45, 0x001b4c},
+ {0x001b80, 0x001b81},
+ {0x001b82, 0x001b82},
+ {0x001b83, 0x001ba0},
+ {0x001ba1, 0x001ba1},
+ {0x001ba2, 0x001ba5},
+ {0x001ba6, 0x001ba7},
+ {0x001ba8, 0x001ba9},
+ {0x001bac, 0x001bad},
+ {0x001bae, 0x001baf},
+ {0x001bba, 0x001be5},
+ {0x001be7, 0x001be7},
+ {0x001be8, 0x001be9},
+ {0x001bea, 0x001bec},
+ {0x001bed, 0x001bed},
+ {0x001bee, 0x001bee},
+ {0x001bef, 0x001bf1},
+ {0x001c00, 0x001c23},
+ {0x001c24, 0x001c2b},
+ {0x001c2c, 0x001c33},
+ {0x001c34, 0x001c35},
+ {0x001c36, 0x001c36},
+ {0x001c4d, 0x001c4f},
+ {0x001c5a, 0x001c77},
+ {0x001c78, 0x001c7d},
+ {0x001c80, 0x001c88},
+ {0x001c90, 0x001cba},
+ {0x001cbd, 0x001cbf},
+ {0x001ce9, 0x001cec},
+ {0x001cee, 0x001cf3},
+ {0x001cf5, 0x001cf6},
+ {0x001cfa, 0x001cfa},
+ {0x001d00, 0x001d2b},
+ {0x001d2c, 0x001d6a},
+ {0x001d6b, 0x001d77},
+ {0x001d78, 0x001d78},
+ {0x001d79, 0x001d9a},
+ {0x001d9b, 0x001dbf},
+ {0x001de7, 0x001df4},
+ {0x001e00, 0x001f15},
+ {0x001f18, 0x001f1d},
+ {0x001f20, 0x001f45},
+ {0x001f48, 0x001f4d},
+ {0x001f50, 0x001f57},
+ {0x001f59, 0x001f59},
+ {0x001f5b, 0x001f5b},
+ {0x001f5d, 0x001f5d},
+ {0x001f5f, 0x001f7d},
+ {0x001f80, 0x001fb4},
+ {0x001fb6, 0x001fbc},
+ {0x001fbe, 0x001fbe},
+ {0x001fc2, 0x001fc4},
+ {0x001fc6, 0x001fcc},
+ {0x001fd0, 0x001fd3},
+ {0x001fd6, 0x001fdb},
+ {0x001fe0, 0x001fec},
+ {0x001ff2, 0x001ff4},
+ {0x001ff6, 0x001ffc},
+ {0x002071, 0x002071},
+ {0x00207f, 0x00207f},
+ {0x002090, 0x00209c},
+ {0x002102, 0x002102},
+ {0x002107, 0x002107},
+ {0x00210a, 0x002113},
+ {0x002115, 0x002115},
+ {0x002119, 0x00211d},
+ {0x002124, 0x002124},
+ {0x002126, 0x002126},
+ {0x002128, 0x002128},
+ {0x00212a, 0x00212d},
+ {0x00212f, 0x002134},
+ {0x002135, 0x002138},
+ {0x002139, 0x002139},
+ {0x00213c, 0x00213f},
+ {0x002145, 0x002149},
+ {0x00214e, 0x00214e},
+ {0x002160, 0x002182},
+ {0x002183, 0x002184},
+ {0x002185, 0x002188},
+ {0x0024b6, 0x0024e9},
+ {0x002c00, 0x002c7b},
+ {0x002c7c, 0x002c7d},
+ {0x002c7e, 0x002ce4},
+ {0x002ceb, 0x002cee},
+ {0x002cf2, 0x002cf3},
+ {0x002d00, 0x002d25},
+ {0x002d27, 0x002d27},
+ {0x002d2d, 0x002d2d},
+ {0x002d30, 0x002d67},
+ {0x002d6f, 0x002d6f},
+ {0x002d80, 0x002d96},
+ {0x002da0, 0x002da6},
+ {0x002da8, 0x002dae},
+ {0x002db0, 0x002db6},
+ {0x002db8, 0x002dbe},
+ {0x002dc0, 0x002dc6},
+ {0x002dc8, 0x002dce},
+ {0x002dd0, 0x002dd6},
+ {0x002dd8, 0x002dde},
+ {0x002de0, 0x002dff},
+ {0x002e2f, 0x002e2f},
+ {0x003005, 0x003005},
+ {0x003006, 0x003006},
+ {0x003007, 0x003007},
+ {0x003021, 0x003029},
+ {0x003031, 0x003035},
+ {0x003038, 0x00303a},
+ {0x00303b, 0x00303b},
+ {0x00303c, 0x00303c},
+ {0x003041, 0x003096},
+ {0x00309d, 0x00309e},
+ {0x00309f, 0x00309f},
+ {0x0030a1, 0x0030fa},
+ {0x0030fc, 0x0030fe},
+ {0x0030ff, 0x0030ff},
+ {0x003105, 0x00312f},
+ {0x003131, 0x00318e},
+ {0x0031a0, 0x0031bf},
+ {0x0031f0, 0x0031ff},
+ {0x003400, 0x004dbf},
+ {0x004e00, 0x00a014},
+ {0x00a015, 0x00a015},
+ {0x00a016, 0x00a48c},
+ {0x00a4d0, 0x00a4f7},
+ {0x00a4f8, 0x00a4fd},
+ {0x00a500, 0x00a60b},
+ {0x00a60c, 0x00a60c},
+ {0x00a610, 0x00a61f},
+ {0x00a62a, 0x00a62b},
+ {0x00a640, 0x00a66d},
+ {0x00a66e, 0x00a66e},
+ {0x00a674, 0x00a67b},
+ {0x00a67f, 0x00a67f},
+ {0x00a680, 0x00a69b},
+ {0x00a69c, 0x00a69d},
+ {0x00a69e, 0x00a69f},
+ {0x00a6a0, 0x00a6e5},
+ {0x00a6e6, 0x00a6ef},
+ {0x00a717, 0x00a71f},
+ {0x00a722, 0x00a76f},
+ {0x00a770, 0x00a770},
+ {0x00a771, 0x00a787},
+ {0x00a788, 0x00a788},
+ {0x00a78b, 0x00a78e},
+ {0x00a78f, 0x00a78f},
+ {0x00a790, 0x00a7ca},
+ {0x00a7d0, 0x00a7d1},
+ {0x00a7d3, 0x00a7d3},
+ {0x00a7d5, 0x00a7d9},
+ {0x00a7f2, 0x00a7f4},
+ {0x00a7f5, 0x00a7f6},
+ {0x00a7f7, 0x00a7f7},
+ {0x00a7f8, 0x00a7f9},
+ {0x00a7fa, 0x00a7fa},
+ {0x00a7fb, 0x00a801},
+ {0x00a802, 0x00a802},
+ {0x00a803, 0x00a805},
+ {0x00a807, 0x00a80a},
+ {0x00a80b, 0x00a80b},
+ {0x00a80c, 0x00a822},
+ {0x00a823, 0x00a824},
+ {0x00a825, 0x00a826},
+ {0x00a827, 0x00a827},
+ {0x00a840, 0x00a873},
+ {0x00a880, 0x00a881},
+ {0x00a882, 0x00a8b3},
+ {0x00a8b4, 0x00a8c3},
+ {0x00a8c5, 0x00a8c5},
+ {0x00a8f2, 0x00a8f7},
+ {0x00a8fb, 0x00a8fb},
+ {0x00a8fd, 0x00a8fe},
+ {0x00a8ff, 0x00a8ff},
+ {0x00a90a, 0x00a925},
+ {0x00a926, 0x00a92a},
+ {0x00a930, 0x00a946},
+ {0x00a947, 0x00a951},
+ {0x00a952, 0x00a952},
+ {0x00a960, 0x00a97c},
+ {0x00a980, 0x00a982},
+ {0x00a983, 0x00a983},
+ {0x00a984, 0x00a9b2},
+ {0x00a9b4, 0x00a9b5},
+ {0x00a9b6, 0x00a9b9},
+ {0x00a9ba, 0x00a9bb},
+ {0x00a9bc, 0x00a9bd},
+ {0x00a9be, 0x00a9bf},
+ {0x00a9cf, 0x00a9cf},
+ {0x00a9e0, 0x00a9e4},
+ {0x00a9e5, 0x00a9e5},
+ {0x00a9e6, 0x00a9e6},
+ {0x00a9e7, 0x00a9ef},
+ {0x00a9fa, 0x00a9fe},
+ {0x00aa00, 0x00aa28},
+ {0x00aa29, 0x00aa2e},
+ {0x00aa2f, 0x00aa30},
+ {0x00aa31, 0x00aa32},
+ {0x00aa33, 0x00aa34},
+ {0x00aa35, 0x00aa36},
+ {0x00aa40, 0x00aa42},
+ {0x00aa43, 0x00aa43},
+ {0x00aa44, 0x00aa4b},
+ {0x00aa4c, 0x00aa4c},
+ {0x00aa4d, 0x00aa4d},
+ {0x00aa60, 0x00aa6f},
+ {0x00aa70, 0x00aa70},
+ {0x00aa71, 0x00aa76},
+ {0x00aa7a, 0x00aa7a},
+ {0x00aa7b, 0x00aa7b},
+ {0x00aa7c, 0x00aa7c},
+ {0x00aa7d, 0x00aa7d},
+ {0x00aa7e, 0x00aaaf},
+ {0x00aab0, 0x00aab0},
+ {0x00aab1, 0x00aab1},
+ {0x00aab2, 0x00aab4},
+ {0x00aab5, 0x00aab6},
+ {0x00aab7, 0x00aab8},
+ {0x00aab9, 0x00aabd},
+ {0x00aabe, 0x00aabe},
+ {0x00aac0, 0x00aac0},
+ {0x00aac2, 0x00aac2},
+ {0x00aadb, 0x00aadc},
+ {0x00aadd, 0x00aadd},
+ {0x00aae0, 0x00aaea},
+ {0x00aaeb, 0x00aaeb},
+ {0x00aaec, 0x00aaed},
+ {0x00aaee, 0x00aaef},
+ {0x00aaf2, 0x00aaf2},
+ {0x00aaf3, 0x00aaf4},
+ {0x00aaf5, 0x00aaf5},
+ {0x00ab01, 0x00ab06},
+ {0x00ab09, 0x00ab0e},
+ {0x00ab11, 0x00ab16},
+ {0x00ab20, 0x00ab26},
+ {0x00ab28, 0x00ab2e},
+ {0x00ab30, 0x00ab5a},
+ {0x00ab5c, 0x00ab5f},
+ {0x00ab60, 0x00ab68},
+ {0x00ab69, 0x00ab69},
+ {0x00ab70, 0x00abbf},
+ {0x00abc0, 0x00abe2},
+ {0x00abe3, 0x00abe4},
+ {0x00abe5, 0x00abe5},
+ {0x00abe6, 0x00abe7},
+ {0x00abe8, 0x00abe8},
+ {0x00abe9, 0x00abea},
+ {0x00ac00, 0x00d7a3},
+ {0x00d7b0, 0x00d7c6},
+ {0x00d7cb, 0x00d7fb},
+ {0x00f900, 0x00fa6d},
+ {0x00fa70, 0x00fad9},
+ {0x00fb00, 0x00fb06},
+ {0x00fb13, 0x00fb17},
+ {0x00fb1d, 0x00fb1d},
+ {0x00fb1e, 0x00fb1e},
+ {0x00fb1f, 0x00fb28},
+ {0x00fb2a, 0x00fb36},
+ {0x00fb38, 0x00fb3c},
+ {0x00fb3e, 0x00fb3e},
+ {0x00fb40, 0x00fb41},
+ {0x00fb43, 0x00fb44},
+ {0x00fb46, 0x00fbb1},
+ {0x00fbd3, 0x00fd3d},
+ {0x00fd50, 0x00fd8f},
+ {0x00fd92, 0x00fdc7},
+ {0x00fdf0, 0x00fdfb},
+ {0x00fe70, 0x00fe74},
+ {0x00fe76, 0x00fefc},
+ {0x00ff21, 0x00ff3a},
+ {0x00ff41, 0x00ff5a},
+ {0x00ff66, 0x00ff6f},
+ {0x00ff70, 0x00ff70},
+ {0x00ff71, 0x00ff9d},
+ {0x00ff9e, 0x00ff9f},
+ {0x00ffa0, 0x00ffbe},
+ {0x00ffc2, 0x00ffc7},
+ {0x00ffca, 0x00ffcf},
+ {0x00ffd2, 0x00ffd7},
+ {0x00ffda, 0x00ffdc},
+ {0x010000, 0x01000b},
+ {0x01000d, 0x010026},
+ {0x010028, 0x01003a},
+ {0x01003c, 0x01003d},
+ {0x01003f, 0x01004d},
+ {0x010050, 0x01005d},
+ {0x010080, 0x0100fa},
+ {0x010140, 0x010174},
+ {0x010280, 0x01029c},
+ {0x0102a0, 0x0102d0},
+ {0x010300, 0x01031f},
+ {0x01032d, 0x010340},
+ {0x010341, 0x010341},
+ {0x010342, 0x010349},
+ {0x01034a, 0x01034a},
+ {0x010350, 0x010375},
+ {0x010376, 0x01037a},
+ {0x010380, 0x01039d},
+ {0x0103a0, 0x0103c3},
+ {0x0103c8, 0x0103cf},
+ {0x0103d1, 0x0103d5},
+ {0x010400, 0x01044f},
+ {0x010450, 0x01049d},
+ {0x0104b0, 0x0104d3},
+ {0x0104d8, 0x0104fb},
+ {0x010500, 0x010527},
+ {0x010530, 0x010563},
+ {0x010570, 0x01057a},
+ {0x01057c, 0x01058a},
+ {0x01058c, 0x010592},
+ {0x010594, 0x010595},
+ {0x010597, 0x0105a1},
+ {0x0105a3, 0x0105b1},
+ {0x0105b3, 0x0105b9},
+ {0x0105bb, 0x0105bc},
+ {0x010600, 0x010736},
+ {0x010740, 0x010755},
+ {0x010760, 0x010767},
+ {0x010780, 0x010785},
+ {0x010787, 0x0107b0},
+ {0x0107b2, 0x0107ba},
+ {0x010800, 0x010805},
+ {0x010808, 0x010808},
+ {0x01080a, 0x010835},
+ {0x010837, 0x010838},
+ {0x01083c, 0x01083c},
+ {0x01083f, 0x010855},
+ {0x010860, 0x010876},
+ {0x010880, 0x01089e},
+ {0x0108e0, 0x0108f2},
+ {0x0108f4, 0x0108f5},
+ {0x010900, 0x010915},
+ {0x010920, 0x010939},
+ {0x010980, 0x0109b7},
+ {0x0109be, 0x0109bf},
+ {0x010a00, 0x010a00},
+ {0x010a01, 0x010a03},
+ {0x010a05, 0x010a06},
+ {0x010a0c, 0x010a0f},
+ {0x010a10, 0x010a13},
+ {0x010a15, 0x010a17},
+ {0x010a19, 0x010a35},
+ {0x010a60, 0x010a7c},
+ {0x010a80, 0x010a9c},
+ {0x010ac0, 0x010ac7},
+ {0x010ac9, 0x010ae4},
+ {0x010b00, 0x010b35},
+ {0x010b40, 0x010b55},
+ {0x010b60, 0x010b72},
+ {0x010b80, 0x010b91},
+ {0x010c00, 0x010c48},
+ {0x010c80, 0x010cb2},
+ {0x010cc0, 0x010cf2},
+ {0x010d00, 0x010d23},
+ {0x010d24, 0x010d27},
+ {0x010e80, 0x010ea9},
+ {0x010eab, 0x010eac},
+ {0x010eb0, 0x010eb1},
+ {0x010f00, 0x010f1c},
+ {0x010f27, 0x010f27},
+ {0x010f30, 0x010f45},
+ {0x010f70, 0x010f81},
+ {0x010fb0, 0x010fc4},
+ {0x010fe0, 0x010ff6},
+ {0x011000, 0x011000},
+ {0x011001, 0x011001},
+ {0x011002, 0x011002},
+ {0x011003, 0x011037},
+ {0x011038, 0x011045},
+ {0x011071, 0x011072},
+ {0x011073, 0x011074},
+ {0x011075, 0x011075},
+ {0x011080, 0x011081},
+ {0x011082, 0x011082},
+ {0x011083, 0x0110af},
+ {0x0110b0, 0x0110b2},
+ {0x0110b3, 0x0110b6},
+ {0x0110b7, 0x0110b8},
+ {0x0110c2, 0x0110c2},
+ {0x0110d0, 0x0110e8},
+ {0x011100, 0x011102},
+ {0x011103, 0x011126},
+ {0x011127, 0x01112b},
+ {0x01112c, 0x01112c},
+ {0x01112d, 0x011132},
+ {0x011144, 0x011144},
+ {0x011145, 0x011146},
+ {0x011147, 0x011147},
+ {0x011150, 0x011172},
+ {0x011176, 0x011176},
+ {0x011180, 0x011181},
+ {0x011182, 0x011182},
+ {0x011183, 0x0111b2},
+ {0x0111b3, 0x0111b5},
+ {0x0111b6, 0x0111be},
+ {0x0111bf, 0x0111bf},
+ {0x0111c1, 0x0111c4},
+ {0x0111ce, 0x0111ce},
+ {0x0111cf, 0x0111cf},
+ {0x0111da, 0x0111da},
+ {0x0111dc, 0x0111dc},
+ {0x011200, 0x011211},
+ {0x011213, 0x01122b},
+ {0x01122c, 0x01122e},
+ {0x01122f, 0x011231},
+ {0x011232, 0x011233},
+ {0x011234, 0x011234},
+ {0x011237, 0x011237},
+ {0x01123e, 0x01123e},
+ {0x01123f, 0x011240},
+ {0x011241, 0x011241},
+ {0x011280, 0x011286},
+ {0x011288, 0x011288},
+ {0x01128a, 0x01128d},
+ {0x01128f, 0x01129d},
+ {0x01129f, 0x0112a8},
+ {0x0112b0, 0x0112de},
+ {0x0112df, 0x0112df},
+ {0x0112e0, 0x0112e2},
+ {0x0112e3, 0x0112e8},
+ {0x011300, 0x011301},
+ {0x011302, 0x011303},
+ {0x011305, 0x01130c},
+ {0x01130f, 0x011310},
+ {0x011313, 0x011328},
+ {0x01132a, 0x011330},
+ {0x011332, 0x011333},
+ {0x011335, 0x011339},
+ {0x01133d, 0x01133d},
+ {0x01133e, 0x01133f},
+ {0x011340, 0x011340},
+ {0x011341, 0x011344},
+ {0x011347, 0x011348},
+ {0x01134b, 0x01134c},
+ {0x011350, 0x011350},
+ {0x011357, 0x011357},
+ {0x01135d, 0x011361},
+ {0x011362, 0x011363},
+ {0x011400, 0x011434},
+ {0x011435, 0x011437},
+ {0x011438, 0x01143f},
+ {0x011440, 0x011441},
+ {0x011443, 0x011444},
+ {0x011445, 0x011445},
+ {0x011447, 0x01144a},
+ {0x01145f, 0x011461},
+ {0x011480, 0x0114af},
+ {0x0114b0, 0x0114b2},
+ {0x0114b3, 0x0114b8},
+ {0x0114b9, 0x0114b9},
+ {0x0114ba, 0x0114ba},
+ {0x0114bb, 0x0114be},
+ {0x0114bf, 0x0114c0},
+ {0x0114c1, 0x0114c1},
+ {0x0114c4, 0x0114c5},
+ {0x0114c7, 0x0114c7},
+ {0x011580, 0x0115ae},
+ {0x0115af, 0x0115b1},
+ {0x0115b2, 0x0115b5},
+ {0x0115b8, 0x0115bb},
+ {0x0115bc, 0x0115bd},
+ {0x0115be, 0x0115be},
+ {0x0115d8, 0x0115db},
+ {0x0115dc, 0x0115dd},
+ {0x011600, 0x01162f},
+ {0x011630, 0x011632},
+ {0x011633, 0x01163a},
+ {0x01163b, 0x01163c},
+ {0x01163d, 0x01163d},
+ {0x01163e, 0x01163e},
+ {0x011640, 0x011640},
+ {0x011644, 0x011644},
+ {0x011680, 0x0116aa},
+ {0x0116ab, 0x0116ab},
+ {0x0116ac, 0x0116ac},
+ {0x0116ad, 0x0116ad},
+ {0x0116ae, 0x0116af},
+ {0x0116b0, 0x0116b5},
+ {0x0116b8, 0x0116b8},
+ {0x011700, 0x01171a},
+ {0x01171d, 0x01171f},
+ {0x011720, 0x011721},
+ {0x011722, 0x011725},
+ {0x011726, 0x011726},
+ {0x011727, 0x01172a},
+ {0x011740, 0x011746},
+ {0x011800, 0x01182b},
+ {0x01182c, 0x01182e},
+ {0x01182f, 0x011837},
+ {0x011838, 0x011838},
+ {0x0118a0, 0x0118df},
+ {0x0118ff, 0x011906},
+ {0x011909, 0x011909},
+ {0x01190c, 0x011913},
+ {0x011915, 0x011916},
+ {0x011918, 0x01192f},
+ {0x011930, 0x011935},
+ {0x011937, 0x011938},
+ {0x01193b, 0x01193c},
+ {0x01193f, 0x01193f},
+ {0x011940, 0x011940},
+ {0x011941, 0x011941},
+ {0x011942, 0x011942},
+ {0x0119a0, 0x0119a7},
+ {0x0119aa, 0x0119d0},
+ {0x0119d1, 0x0119d3},
+ {0x0119d4, 0x0119d7},
+ {0x0119da, 0x0119db},
+ {0x0119dc, 0x0119df},
+ {0x0119e1, 0x0119e1},
+ {0x0119e3, 0x0119e3},
+ {0x0119e4, 0x0119e4},
+ {0x011a00, 0x011a00},
+ {0x011a01, 0x011a0a},
+ {0x011a0b, 0x011a32},
+ {0x011a35, 0x011a38},
+ {0x011a39, 0x011a39},
+ {0x011a3a, 0x011a3a},
+ {0x011a3b, 0x011a3e},
+ {0x011a50, 0x011a50},
+ {0x011a51, 0x011a56},
+ {0x011a57, 0x011a58},
+ {0x011a59, 0x011a5b},
+ {0x011a5c, 0x011a89},
+ {0x011a8a, 0x011a96},
+ {0x011a97, 0x011a97},
+ {0x011a9d, 0x011a9d},
+ {0x011ab0, 0x011af8},
+ {0x011c00, 0x011c08},
+ {0x011c0a, 0x011c2e},
+ {0x011c2f, 0x011c2f},
+ {0x011c30, 0x011c36},
+ {0x011c38, 0x011c3d},
+ {0x011c3e, 0x011c3e},
+ {0x011c40, 0x011c40},
+ {0x011c72, 0x011c8f},
+ {0x011c92, 0x011ca7},
+ {0x011ca9, 0x011ca9},
+ {0x011caa, 0x011cb0},
+ {0x011cb1, 0x011cb1},
+ {0x011cb2, 0x011cb3},
+ {0x011cb4, 0x011cb4},
+ {0x011cb5, 0x011cb6},
+ {0x011d00, 0x011d06},
+ {0x011d08, 0x011d09},
+ {0x011d0b, 0x011d30},
+ {0x011d31, 0x011d36},
+ {0x011d3a, 0x011d3a},
+ {0x011d3c, 0x011d3d},
+ {0x011d3f, 0x011d41},
+ {0x011d43, 0x011d43},
+ {0x011d46, 0x011d46},
+ {0x011d47, 0x011d47},
+ {0x011d60, 0x011d65},
+ {0x011d67, 0x011d68},
+ {0x011d6a, 0x011d89},
+ {0x011d8a, 0x011d8e},
+ {0x011d90, 0x011d91},
+ {0x011d93, 0x011d94},
+ {0x011d95, 0x011d95},
+ {0x011d96, 0x011d96},
+ {0x011d98, 0x011d98},
+ {0x011ee0, 0x011ef2},
+ {0x011ef3, 0x011ef4},
+ {0x011ef5, 0x011ef6},
+ {0x011f00, 0x011f01},
+ {0x011f02, 0x011f02},
+ {0x011f03, 0x011f03},
+ {0x011f04, 0x011f10},
+ {0x011f12, 0x011f33},
+ {0x011f34, 0x011f35},
+ {0x011f36, 0x011f3a},
+ {0x011f3e, 0x011f3f},
+ {0x011f40, 0x011f40},
+ {0x011fb0, 0x011fb0},
+ {0x012000, 0x012399},
+ {0x012400, 0x01246e},
+ {0x012480, 0x012543},
+ {0x012f90, 0x012ff0},
+ {0x013000, 0x01342f},
+ {0x013441, 0x013446},
+ {0x014400, 0x014646},
+ {0x016800, 0x016a38},
+ {0x016a40, 0x016a5e},
+ {0x016a70, 0x016abe},
+ {0x016ad0, 0x016aed},
+ {0x016b00, 0x016b2f},
+ {0x016b40, 0x016b43},
+ {0x016b63, 0x016b77},
+ {0x016b7d, 0x016b8f},
+ {0x016e40, 0x016e7f},
+ {0x016f00, 0x016f4a},
+ {0x016f4f, 0x016f4f},
+ {0x016f50, 0x016f50},
+ {0x016f51, 0x016f87},
+ {0x016f8f, 0x016f92},
+ {0x016f93, 0x016f9f},
+ {0x016fe0, 0x016fe1},
+ {0x016fe3, 0x016fe3},
+ {0x016ff0, 0x016ff1},
+ {0x017000, 0x0187f7},
+ {0x018800, 0x018cd5},
+ {0x018d00, 0x018d08},
+ {0x01aff0, 0x01aff3},
+ {0x01aff5, 0x01affb},
+ {0x01affd, 0x01affe},
+ {0x01b000, 0x01b122},
+ {0x01b132, 0x01b132},
+ {0x01b150, 0x01b152},
+ {0x01b155, 0x01b155},
+ {0x01b164, 0x01b167},
+ {0x01b170, 0x01b2fb},
+ {0x01bc00, 0x01bc6a},
+ {0x01bc70, 0x01bc7c},
+ {0x01bc80, 0x01bc88},
+ {0x01bc90, 0x01bc99},
+ {0x01bc9e, 0x01bc9e},
+ {0x01d400, 0x01d454},
+ {0x01d456, 0x01d49c},
+ {0x01d49e, 0x01d49f},
+ {0x01d4a2, 0x01d4a2},
+ {0x01d4a5, 0x01d4a6},
+ {0x01d4a9, 0x01d4ac},
+ {0x01d4ae, 0x01d4b9},
+ {0x01d4bb, 0x01d4bb},
+ {0x01d4bd, 0x01d4c3},
+ {0x01d4c5, 0x01d505},
+ {0x01d507, 0x01d50a},
+ {0x01d50d, 0x01d514},
+ {0x01d516, 0x01d51c},
+ {0x01d51e, 0x01d539},
+ {0x01d53b, 0x01d53e},
+ {0x01d540, 0x01d544},
+ {0x01d546, 0x01d546},
+ {0x01d54a, 0x01d550},
+ {0x01d552, 0x01d6a5},
+ {0x01d6a8, 0x01d6c0},
+ {0x01d6c2, 0x01d6da},
+ {0x01d6dc, 0x01d6fa},
+ {0x01d6fc, 0x01d714},
+ {0x01d716, 0x01d734},
+ {0x01d736, 0x01d74e},
+ {0x01d750, 0x01d76e},
+ {0x01d770, 0x01d788},
+ {0x01d78a, 0x01d7a8},
+ {0x01d7aa, 0x01d7c2},
+ {0x01d7c4, 0x01d7cb},
+ {0x01df00, 0x01df09},
+ {0x01df0a, 0x01df0a},
+ {0x01df0b, 0x01df1e},
+ {0x01df25, 0x01df2a},
+ {0x01e000, 0x01e006},
+ {0x01e008, 0x01e018},
+ {0x01e01b, 0x01e021},
+ {0x01e023, 0x01e024},
+ {0x01e026, 0x01e02a},
+ {0x01e030, 0x01e06d},
+ {0x01e08f, 0x01e08f},
+ {0x01e100, 0x01e12c},
+ {0x01e137, 0x01e13d},
+ {0x01e14e, 0x01e14e},
+ {0x01e290, 0x01e2ad},
+ {0x01e2c0, 0x01e2eb},
+ {0x01e4d0, 0x01e4ea},
+ {0x01e4eb, 0x01e4eb},
+ {0x01e7e0, 0x01e7e6},
+ {0x01e7e8, 0x01e7eb},
+ {0x01e7ed, 0x01e7ee},
+ {0x01e7f0, 0x01e7fe},
+ {0x01e800, 0x01e8c4},
+ {0x01e900, 0x01e943},
+ {0x01e947, 0x01e947},
+ {0x01e94b, 0x01e94b},
+ {0x01ee00, 0x01ee03},
+ {0x01ee05, 0x01ee1f},
+ {0x01ee21, 0x01ee22},
+ {0x01ee24, 0x01ee24},
+ {0x01ee27, 0x01ee27},
+ {0x01ee29, 0x01ee32},
+ {0x01ee34, 0x01ee37},
+ {0x01ee39, 0x01ee39},
+ {0x01ee3b, 0x01ee3b},
+ {0x01ee42, 0x01ee42},
+ {0x01ee47, 0x01ee47},
+ {0x01ee49, 0x01ee49},
+ {0x01ee4b, 0x01ee4b},
+ {0x01ee4d, 0x01ee4f},
+ {0x01ee51, 0x01ee52},
+ {0x01ee54, 0x01ee54},
+ {0x01ee57, 0x01ee57},
+ {0x01ee59, 0x01ee59},
+ {0x01ee5b, 0x01ee5b},
+ {0x01ee5d, 0x01ee5d},
+ {0x01ee5f, 0x01ee5f},
+ {0x01ee61, 0x01ee62},
+ {0x01ee64, 0x01ee64},
+ {0x01ee67, 0x01ee6a},
+ {0x01ee6c, 0x01ee72},
+ {0x01ee74, 0x01ee77},
+ {0x01ee79, 0x01ee7c},
+ {0x01ee7e, 0x01ee7e},
+ {0x01ee80, 0x01ee89},
+ {0x01ee8b, 0x01ee9b},
+ {0x01eea1, 0x01eea3},
+ {0x01eea5, 0x01eea9},
+ {0x01eeab, 0x01eebb},
+ {0x01f130, 0x01f149},
+ {0x01f150, 0x01f169},
+ {0x01f170, 0x01f189},
+ {0x020000, 0x02a6df},
+ {0x02a700, 0x02b739},
+ {0x02b740, 0x02b81d},
+ {0x02b820, 0x02cea1},
+ {0x02ceb0, 0x02ebe0},
+ {0x02ebf0, 0x02ee5d},
+ {0x02f800, 0x02fa1d},
+ {0x030000, 0x03134a},
+ {0x031350, 0x0323af}
+};
+
+/* table of Unicode codepoint ranges of Lowercase characters */
+static const pg_unicode_range unicode_lowercase[686] =
+{
+ {0x000061, 0x00007a},
+ {0x0000aa, 0x0000aa},
+ {0x0000b5, 0x0000b5},
+ {0x0000ba, 0x0000ba},
+ {0x0000df, 0x0000f6},
+ {0x0000f8, 0x0000ff},
+ {0x000101, 0x000101},
+ {0x000103, 0x000103},
+ {0x000105, 0x000105},
+ {0x000107, 0x000107},
+ {0x000109, 0x000109},
+ {0x00010b, 0x00010b},
+ {0x00010d, 0x00010d},
+ {0x00010f, 0x00010f},
+ {0x000111, 0x000111},
+ {0x000113, 0x000113},
+ {0x000115, 0x000115},
+ {0x000117, 0x000117},
+ {0x000119, 0x000119},
+ {0x00011b, 0x00011b},
+ {0x00011d, 0x00011d},
+ {0x00011f, 0x00011f},
+ {0x000121, 0x000121},
+ {0x000123, 0x000123},
+ {0x000125, 0x000125},
+ {0x000127, 0x000127},
+ {0x000129, 0x000129},
+ {0x00012b, 0x00012b},
+ {0x00012d, 0x00012d},
+ {0x00012f, 0x00012f},
+ {0x000131, 0x000131},
+ {0x000133, 0x000133},
+ {0x000135, 0x000135},
+ {0x000137, 0x000138},
+ {0x00013a, 0x00013a},
+ {0x00013c, 0x00013c},
+ {0x00013e, 0x00013e},
+ {0x000140, 0x000140},
+ {0x000142, 0x000142},
+ {0x000144, 0x000144},
+ {0x000146, 0x000146},
+ {0x000148, 0x000149},
+ {0x00014b, 0x00014b},
+ {0x00014d, 0x00014d},
+ {0x00014f, 0x00014f},
+ {0x000151, 0x000151},
+ {0x000153, 0x000153},
+ {0x000155, 0x000155},
+ {0x000157, 0x000157},
+ {0x000159, 0x000159},
+ {0x00015b, 0x00015b},
+ {0x00015d, 0x00015d},
+ {0x00015f, 0x00015f},
+ {0x000161, 0x000161},
+ {0x000163, 0x000163},
+ {0x000165, 0x000165},
+ {0x000167, 0x000167},
+ {0x000169, 0x000169},
+ {0x00016b, 0x00016b},
+ {0x00016d, 0x00016d},
+ {0x00016f, 0x00016f},
+ {0x000171, 0x000171},
+ {0x000173, 0x000173},
+ {0x000175, 0x000175},
+ {0x000177, 0x000177},
+ {0x00017a, 0x00017a},
+ {0x00017c, 0x00017c},
+ {0x00017e, 0x000180},
+ {0x000183, 0x000183},
+ {0x000185, 0x000185},
+ {0x000188, 0x000188},
+ {0x00018c, 0x00018d},
+ {0x000192, 0x000192},
+ {0x000195, 0x000195},
+ {0x000199, 0x00019b},
+ {0x00019e, 0x00019e},
+ {0x0001a1, 0x0001a1},
+ {0x0001a3, 0x0001a3},
+ {0x0001a5, 0x0001a5},
+ {0x0001a8, 0x0001a8},
+ {0x0001aa, 0x0001ab},
+ {0x0001ad, 0x0001ad},
+ {0x0001b0, 0x0001b0},
+ {0x0001b4, 0x0001b4},
+ {0x0001b6, 0x0001b6},
+ {0x0001b9, 0x0001ba},
+ {0x0001bd, 0x0001bf},
+ {0x0001c6, 0x0001c6},
+ {0x0001c9, 0x0001c9},
+ {0x0001cc, 0x0001cc},
+ {0x0001ce, 0x0001ce},
+ {0x0001d0, 0x0001d0},
+ {0x0001d2, 0x0001d2},
+ {0x0001d4, 0x0001d4},
+ {0x0001d6, 0x0001d6},
+ {0x0001d8, 0x0001d8},
+ {0x0001da, 0x0001da},
+ {0x0001dc, 0x0001dd},
+ {0x0001df, 0x0001df},
+ {0x0001e1, 0x0001e1},
+ {0x0001e3, 0x0001e3},
+ {0x0001e5, 0x0001e5},
+ {0x0001e7, 0x0001e7},
+ {0x0001e9, 0x0001e9},
+ {0x0001eb, 0x0001eb},
+ {0x0001ed, 0x0001ed},
+ {0x0001ef, 0x0001f0},
+ {0x0001f3, 0x0001f3},
+ {0x0001f5, 0x0001f5},
+ {0x0001f9, 0x0001f9},
+ {0x0001fb, 0x0001fb},
+ {0x0001fd, 0x0001fd},
+ {0x0001ff, 0x0001ff},
+ {0x000201, 0x000201},
+ {0x000203, 0x000203},
+ {0x000205, 0x000205},
+ {0x000207, 0x000207},
+ {0x000209, 0x000209},
+ {0x00020b, 0x00020b},
+ {0x00020d, 0x00020d},
+ {0x00020f, 0x00020f},
+ {0x000211, 0x000211},
+ {0x000213, 0x000213},
+ {0x000215, 0x000215},
+ {0x000217, 0x000217},
+ {0x000219, 0x000219},
+ {0x00021b, 0x00021b},
+ {0x00021d, 0x00021d},
+ {0x00021f, 0x00021f},
+ {0x000221, 0x000221},
+ {0x000223, 0x000223},
+ {0x000225, 0x000225},
+ {0x000227, 0x000227},
+ {0x000229, 0x000229},
+ {0x00022b, 0x00022b},
+ {0x00022d, 0x00022d},
+ {0x00022f, 0x00022f},
+ {0x000231, 0x000231},
+ {0x000233, 0x000239},
+ {0x00023c, 0x00023c},
+ {0x00023f, 0x000240},
+ {0x000242, 0x000242},
+ {0x000247, 0x000247},
+ {0x000249, 0x000249},
+ {0x00024b, 0x00024b},
+ {0x00024d, 0x00024d},
+ {0x00024f, 0x000293},
+ {0x000295, 0x0002af},
+ {0x0002b0, 0x0002b8},
+ {0x0002c0, 0x0002c1},
+ {0x0002e0, 0x0002e4},
+ {0x000345, 0x000345},
+ {0x000371, 0x000371},
+ {0x000373, 0x000373},
+ {0x000377, 0x000377},
+ {0x00037a, 0x00037a},
+ {0x00037b, 0x00037d},
+ {0x000390, 0x000390},
+ {0x0003ac, 0x0003ce},
+ {0x0003d0, 0x0003d1},
+ {0x0003d5, 0x0003d7},
+ {0x0003d9, 0x0003d9},
+ {0x0003db, 0x0003db},
+ {0x0003dd, 0x0003dd},
+ {0x0003df, 0x0003df},
+ {0x0003e1, 0x0003e1},
+ {0x0003e3, 0x0003e3},
+ {0x0003e5, 0x0003e5},
+ {0x0003e7, 0x0003e7},
+ {0x0003e9, 0x0003e9},
+ {0x0003eb, 0x0003eb},
+ {0x0003ed, 0x0003ed},
+ {0x0003ef, 0x0003f3},
+ {0x0003f5, 0x0003f5},
+ {0x0003f8, 0x0003f8},
+ {0x0003fb, 0x0003fc},
+ {0x000430, 0x00045f},
+ {0x000461, 0x000461},
+ {0x000463, 0x000463},
+ {0x000465, 0x000465},
+ {0x000467, 0x000467},
+ {0x000469, 0x000469},
+ {0x00046b, 0x00046b},
+ {0x00046d, 0x00046d},
+ {0x00046f, 0x00046f},
+ {0x000471, 0x000471},
+ {0x000473, 0x000473},
+ {0x000475, 0x000475},
+ {0x000477, 0x000477},
+ {0x000479, 0x000479},
+ {0x00047b, 0x00047b},
+ {0x00047d, 0x00047d},
+ {0x00047f, 0x00047f},
+ {0x000481, 0x000481},
+ {0x00048b, 0x00048b},
+ {0x00048d, 0x00048d},
+ {0x00048f, 0x00048f},
+ {0x000491, 0x000491},
+ {0x000493, 0x000493},
+ {0x000495, 0x000495},
+ {0x000497, 0x000497},
+ {0x000499, 0x000499},
+ {0x00049b, 0x00049b},
+ {0x00049d, 0x00049d},
+ {0x00049f, 0x00049f},
+ {0x0004a1, 0x0004a1},
+ {0x0004a3, 0x0004a3},
+ {0x0004a5, 0x0004a5},
+ {0x0004a7, 0x0004a7},
+ {0x0004a9, 0x0004a9},
+ {0x0004ab, 0x0004ab},
+ {0x0004ad, 0x0004ad},
+ {0x0004af, 0x0004af},
+ {0x0004b1, 0x0004b1},
+ {0x0004b3, 0x0004b3},
+ {0x0004b5, 0x0004b5},
+ {0x0004b7, 0x0004b7},
+ {0x0004b9, 0x0004b9},
+ {0x0004bb, 0x0004bb},
+ {0x0004bd, 0x0004bd},
+ {0x0004bf, 0x0004bf},
+ {0x0004c2, 0x0004c2},
+ {0x0004c4, 0x0004c4},
+ {0x0004c6, 0x0004c6},
+ {0x0004c8, 0x0004c8},
+ {0x0004ca, 0x0004ca},
+ {0x0004cc, 0x0004cc},
+ {0x0004ce, 0x0004cf},
+ {0x0004d1, 0x0004d1},
+ {0x0004d3, 0x0004d3},
+ {0x0004d5, 0x0004d5},
+ {0x0004d7, 0x0004d7},
+ {0x0004d9, 0x0004d9},
+ {0x0004db, 0x0004db},
+ {0x0004dd, 0x0004dd},
+ {0x0004df, 0x0004df},
+ {0x0004e1, 0x0004e1},
+ {0x0004e3, 0x0004e3},
+ {0x0004e5, 0x0004e5},
+ {0x0004e7, 0x0004e7},
+ {0x0004e9, 0x0004e9},
+ {0x0004eb, 0x0004eb},
+ {0x0004ed, 0x0004ed},
+ {0x0004ef, 0x0004ef},
+ {0x0004f1, 0x0004f1},
+ {0x0004f3, 0x0004f3},
+ {0x0004f5, 0x0004f5},
+ {0x0004f7, 0x0004f7},
+ {0x0004f9, 0x0004f9},
+ {0x0004fb, 0x0004fb},
+ {0x0004fd, 0x0004fd},
+ {0x0004ff, 0x0004ff},
+ {0x000501, 0x000501},
+ {0x000503, 0x000503},
+ {0x000505, 0x000505},
+ {0x000507, 0x000507},
+ {0x000509, 0x000509},
+ {0x00050b, 0x00050b},
+ {0x00050d, 0x00050d},
+ {0x00050f, 0x00050f},
+ {0x000511, 0x000511},
+ {0x000513, 0x000513},
+ {0x000515, 0x000515},
+ {0x000517, 0x000517},
+ {0x000519, 0x000519},
+ {0x00051b, 0x00051b},
+ {0x00051d, 0x00051d},
+ {0x00051f, 0x00051f},
+ {0x000521, 0x000521},
+ {0x000523, 0x000523},
+ {0x000525, 0x000525},
+ {0x000527, 0x000527},
+ {0x000529, 0x000529},
+ {0x00052b, 0x00052b},
+ {0x00052d, 0x00052d},
+ {0x00052f, 0x00052f},
+ {0x000560, 0x000588},
+ {0x0010d0, 0x0010fa},
+ {0x0010fc, 0x0010fc},
+ {0x0010fd, 0x0010ff},
+ {0x0013f8, 0x0013fd},
+ {0x001c80, 0x001c88},
+ {0x001d00, 0x001d2b},
+ {0x001d2c, 0x001d6a},
+ {0x001d6b, 0x001d77},
+ {0x001d78, 0x001d78},
+ {0x001d79, 0x001d9a},
+ {0x001d9b, 0x001dbf},
+ {0x001e01, 0x001e01},
+ {0x001e03, 0x001e03},
+ {0x001e05, 0x001e05},
+ {0x001e07, 0x001e07},
+ {0x001e09, 0x001e09},
+ {0x001e0b, 0x001e0b},
+ {0x001e0d, 0x001e0d},
+ {0x001e0f, 0x001e0f},
+ {0x001e11, 0x001e11},
+ {0x001e13, 0x001e13},
+ {0x001e15, 0x001e15},
+ {0x001e17, 0x001e17},
+ {0x001e19, 0x001e19},
+ {0x001e1b, 0x001e1b},
+ {0x001e1d, 0x001e1d},
+ {0x001e1f, 0x001e1f},
+ {0x001e21, 0x001e21},
+ {0x001e23, 0x001e23},
+ {0x001e25, 0x001e25},
+ {0x001e27, 0x001e27},
+ {0x001e29, 0x001e29},
+ {0x001e2b, 0x001e2b},
+ {0x001e2d, 0x001e2d},
+ {0x001e2f, 0x001e2f},
+ {0x001e31, 0x001e31},
+ {0x001e33, 0x001e33},
+ {0x001e35, 0x001e35},
+ {0x001e37, 0x001e37},
+ {0x001e39, 0x001e39},
+ {0x001e3b, 0x001e3b},
+ {0x001e3d, 0x001e3d},
+ {0x001e3f, 0x001e3f},
+ {0x001e41, 0x001e41},
+ {0x001e43, 0x001e43},
+ {0x001e45, 0x001e45},
+ {0x001e47, 0x001e47},
+ {0x001e49, 0x001e49},
+ {0x001e4b, 0x001e4b},
+ {0x001e4d, 0x001e4d},
+ {0x001e4f, 0x001e4f},
+ {0x001e51, 0x001e51},
+ {0x001e53, 0x001e53},
+ {0x001e55, 0x001e55},
+ {0x001e57, 0x001e57},
+ {0x001e59, 0x001e59},
+ {0x001e5b, 0x001e5b},
+ {0x001e5d, 0x001e5d},
+ {0x001e5f, 0x001e5f},
+ {0x001e61, 0x001e61},
+ {0x001e63, 0x001e63},
+ {0x001e65, 0x001e65},
+ {0x001e67, 0x001e67},
+ {0x001e69, 0x001e69},
+ {0x001e6b, 0x001e6b},
+ {0x001e6d, 0x001e6d},
+ {0x001e6f, 0x001e6f},
+ {0x001e71, 0x001e71},
+ {0x001e73, 0x001e73},
+ {0x001e75, 0x001e75},
+ {0x001e77, 0x001e77},
+ {0x001e79, 0x001e79},
+ {0x001e7b, 0x001e7b},
+ {0x001e7d, 0x001e7d},
+ {0x001e7f, 0x001e7f},
+ {0x001e81, 0x001e81},
+ {0x001e83, 0x001e83},
+ {0x001e85, 0x001e85},
+ {0x001e87, 0x001e87},
+ {0x001e89, 0x001e89},
+ {0x001e8b, 0x001e8b},
+ {0x001e8d, 0x001e8d},
+ {0x001e8f, 0x001e8f},
+ {0x001e91, 0x001e91},
+ {0x001e93, 0x001e93},
+ {0x001e95, 0x001e9d},
+ {0x001e9f, 0x001e9f},
+ {0x001ea1, 0x001ea1},
+ {0x001ea3, 0x001ea3},
+ {0x001ea5, 0x001ea5},
+ {0x001ea7, 0x001ea7},
+ {0x001ea9, 0x001ea9},
+ {0x001eab, 0x001eab},
+ {0x001ead, 0x001ead},
+ {0x001eaf, 0x001eaf},
+ {0x001eb1, 0x001eb1},
+ {0x001eb3, 0x001eb3},
+ {0x001eb5, 0x001eb5},
+ {0x001eb7, 0x001eb7},
+ {0x001eb9, 0x001eb9},
+ {0x001ebb, 0x001ebb},
+ {0x001ebd, 0x001ebd},
+ {0x001ebf, 0x001ebf},
+ {0x001ec1, 0x001ec1},
+ {0x001ec3, 0x001ec3},
+ {0x001ec5, 0x001ec5},
+ {0x001ec7, 0x001ec7},
+ {0x001ec9, 0x001ec9},
+ {0x001ecb, 0x001ecb},
+ {0x001ecd, 0x001ecd},
+ {0x001ecf, 0x001ecf},
+ {0x001ed1, 0x001ed1},
+ {0x001ed3, 0x001ed3},
+ {0x001ed5, 0x001ed5},
+ {0x001ed7, 0x001ed7},
+ {0x001ed9, 0x001ed9},
+ {0x001edb, 0x001edb},
+ {0x001edd, 0x001edd},
+ {0x001edf, 0x001edf},
+ {0x001ee1, 0x001ee1},
+ {0x001ee3, 0x001ee3},
+ {0x001ee5, 0x001ee5},
+ {0x001ee7, 0x001ee7},
+ {0x001ee9, 0x001ee9},
+ {0x001eeb, 0x001eeb},
+ {0x001eed, 0x001eed},
+ {0x001eef, 0x001eef},
+ {0x001ef1, 0x001ef1},
+ {0x001ef3, 0x001ef3},
+ {0x001ef5, 0x001ef5},
+ {0x001ef7, 0x001ef7},
+ {0x001ef9, 0x001ef9},
+ {0x001efb, 0x001efb},
+ {0x001efd, 0x001efd},
+ {0x001eff, 0x001f07},
+ {0x001f10, 0x001f15},
+ {0x001f20, 0x001f27},
+ {0x001f30, 0x001f37},
+ {0x001f40, 0x001f45},
+ {0x001f50, 0x001f57},
+ {0x001f60, 0x001f67},
+ {0x001f70, 0x001f7d},
+ {0x001f80, 0x001f87},
+ {0x001f90, 0x001f97},
+ {0x001fa0, 0x001fa7},
+ {0x001fb0, 0x001fb4},
+ {0x001fb6, 0x001fb7},
+ {0x001fbe, 0x001fbe},
+ {0x001fc2, 0x001fc4},
+ {0x001fc6, 0x001fc7},
+ {0x001fd0, 0x001fd3},
+ {0x001fd6, 0x001fd7},
+ {0x001fe0, 0x001fe7},
+ {0x001ff2, 0x001ff4},
+ {0x001ff6, 0x001ff7},
+ {0x002071, 0x002071},
+ {0x00207f, 0x00207f},
+ {0x002090, 0x00209c},
+ {0x00210a, 0x00210a},
+ {0x00210e, 0x00210f},
+ {0x002113, 0x002113},
+ {0x00212f, 0x00212f},
+ {0x002134, 0x002134},
+ {0x002139, 0x002139},
+ {0x00213c, 0x00213d},
+ {0x002146, 0x002149},
+ {0x00214e, 0x00214e},
+ {0x002170, 0x00217f},
+ {0x002184, 0x002184},
+ {0x0024d0, 0x0024e9},
+ {0x002c30, 0x002c5f},
+ {0x002c61, 0x002c61},
+ {0x002c65, 0x002c66},
+ {0x002c68, 0x002c68},
+ {0x002c6a, 0x002c6a},
+ {0x002c6c, 0x002c6c},
+ {0x002c71, 0x002c71},
+ {0x002c73, 0x002c74},
+ {0x002c76, 0x002c7b},
+ {0x002c7c, 0x002c7d},
+ {0x002c81, 0x002c81},
+ {0x002c83, 0x002c83},
+ {0x002c85, 0x002c85},
+ {0x002c87, 0x002c87},
+ {0x002c89, 0x002c89},
+ {0x002c8b, 0x002c8b},
+ {0x002c8d, 0x002c8d},
+ {0x002c8f, 0x002c8f},
+ {0x002c91, 0x002c91},
+ {0x002c93, 0x002c93},
+ {0x002c95, 0x002c95},
+ {0x002c97, 0x002c97},
+ {0x002c99, 0x002c99},
+ {0x002c9b, 0x002c9b},
+ {0x002c9d, 0x002c9d},
+ {0x002c9f, 0x002c9f},
+ {0x002ca1, 0x002ca1},
+ {0x002ca3, 0x002ca3},
+ {0x002ca5, 0x002ca5},
+ {0x002ca7, 0x002ca7},
+ {0x002ca9, 0x002ca9},
+ {0x002cab, 0x002cab},
+ {0x002cad, 0x002cad},
+ {0x002caf, 0x002caf},
+ {0x002cb1, 0x002cb1},
+ {0x002cb3, 0x002cb3},
+ {0x002cb5, 0x002cb5},
+ {0x002cb7, 0x002cb7},
+ {0x002cb9, 0x002cb9},
+ {0x002cbb, 0x002cbb},
+ {0x002cbd, 0x002cbd},
+ {0x002cbf, 0x002cbf},
+ {0x002cc1, 0x002cc1},
+ {0x002cc3, 0x002cc3},
+ {0x002cc5, 0x002cc5},
+ {0x002cc7, 0x002cc7},
+ {0x002cc9, 0x002cc9},
+ {0x002ccb, 0x002ccb},
+ {0x002ccd, 0x002ccd},
+ {0x002ccf, 0x002ccf},
+ {0x002cd1, 0x002cd1},
+ {0x002cd3, 0x002cd3},
+ {0x002cd5, 0x002cd5},
+ {0x002cd7, 0x002cd7},
+ {0x002cd9, 0x002cd9},
+ {0x002cdb, 0x002cdb},
+ {0x002cdd, 0x002cdd},
+ {0x002cdf, 0x002cdf},
+ {0x002ce1, 0x002ce1},
+ {0x002ce3, 0x002ce4},
+ {0x002cec, 0x002cec},
+ {0x002cee, 0x002cee},
+ {0x002cf3, 0x002cf3},
+ {0x002d00, 0x002d25},
+ {0x002d27, 0x002d27},
+ {0x002d2d, 0x002d2d},
+ {0x00a641, 0x00a641},
+ {0x00a643, 0x00a643},
+ {0x00a645, 0x00a645},
+ {0x00a647, 0x00a647},
+ {0x00a649, 0x00a649},
+ {0x00a64b, 0x00a64b},
+ {0x00a64d, 0x00a64d},
+ {0x00a64f, 0x00a64f},
+ {0x00a651, 0x00a651},
+ {0x00a653, 0x00a653},
+ {0x00a655, 0x00a655},
+ {0x00a657, 0x00a657},
+ {0x00a659, 0x00a659},
+ {0x00a65b, 0x00a65b},
+ {0x00a65d, 0x00a65d},
+ {0x00a65f, 0x00a65f},
+ {0x00a661, 0x00a661},
+ {0x00a663, 0x00a663},
+ {0x00a665, 0x00a665},
+ {0x00a667, 0x00a667},
+ {0x00a669, 0x00a669},
+ {0x00a66b, 0x00a66b},
+ {0x00a66d, 0x00a66d},
+ {0x00a681, 0x00a681},
+ {0x00a683, 0x00a683},
+ {0x00a685, 0x00a685},
+ {0x00a687, 0x00a687},
+ {0x00a689, 0x00a689},
+ {0x00a68b, 0x00a68b},
+ {0x00a68d, 0x00a68d},
+ {0x00a68f, 0x00a68f},
+ {0x00a691, 0x00a691},
+ {0x00a693, 0x00a693},
+ {0x00a695, 0x00a695},
+ {0x00a697, 0x00a697},
+ {0x00a699, 0x00a699},
+ {0x00a69b, 0x00a69b},
+ {0x00a69c, 0x00a69d},
+ {0x00a723, 0x00a723},
+ {0x00a725, 0x00a725},
+ {0x00a727, 0x00a727},
+ {0x00a729, 0x00a729},
+ {0x00a72b, 0x00a72b},
+ {0x00a72d, 0x00a72d},
+ {0x00a72f, 0x00a731},
+ {0x00a733, 0x00a733},
+ {0x00a735, 0x00a735},
+ {0x00a737, 0x00a737},
+ {0x00a739, 0x00a739},
+ {0x00a73b, 0x00a73b},
+ {0x00a73d, 0x00a73d},
+ {0x00a73f, 0x00a73f},
+ {0x00a741, 0x00a741},
+ {0x00a743, 0x00a743},
+ {0x00a745, 0x00a745},
+ {0x00a747, 0x00a747},
+ {0x00a749, 0x00a749},
+ {0x00a74b, 0x00a74b},
+ {0x00a74d, 0x00a74d},
+ {0x00a74f, 0x00a74f},
+ {0x00a751, 0x00a751},
+ {0x00a753, 0x00a753},
+ {0x00a755, 0x00a755},
+ {0x00a757, 0x00a757},
+ {0x00a759, 0x00a759},
+ {0x00a75b, 0x00a75b},
+ {0x00a75d, 0x00a75d},
+ {0x00a75f, 0x00a75f},
+ {0x00a761, 0x00a761},
+ {0x00a763, 0x00a763},
+ {0x00a765, 0x00a765},
+ {0x00a767, 0x00a767},
+ {0x00a769, 0x00a769},
+ {0x00a76b, 0x00a76b},
+ {0x00a76d, 0x00a76d},
+ {0x00a76f, 0x00a76f},
+ {0x00a770, 0x00a770},
+ {0x00a771, 0x00a778},
+ {0x00a77a, 0x00a77a},
+ {0x00a77c, 0x00a77c},
+ {0x00a77f, 0x00a77f},
+ {0x00a781, 0x00a781},
+ {0x00a783, 0x00a783},
+ {0x00a785, 0x00a785},
+ {0x00a787, 0x00a787},
+ {0x00a78c, 0x00a78c},
+ {0x00a78e, 0x00a78e},
+ {0x00a791, 0x00a791},
+ {0x00a793, 0x00a795},
+ {0x00a797, 0x00a797},
+ {0x00a799, 0x00a799},
+ {0x00a79b, 0x00a79b},
+ {0x00a79d, 0x00a79d},
+ {0x00a79f, 0x00a79f},
+ {0x00a7a1, 0x00a7a1},
+ {0x00a7a3, 0x00a7a3},
+ {0x00a7a5, 0x00a7a5},
+ {0x00a7a7, 0x00a7a7},
+ {0x00a7a9, 0x00a7a9},
+ {0x00a7af, 0x00a7af},
+ {0x00a7b5, 0x00a7b5},
+ {0x00a7b7, 0x00a7b7},
+ {0x00a7b9, 0x00a7b9},
+ {0x00a7bb, 0x00a7bb},
+ {0x00a7bd, 0x00a7bd},
+ {0x00a7bf, 0x00a7bf},
+ {0x00a7c1, 0x00a7c1},
+ {0x00a7c3, 0x00a7c3},
+ {0x00a7c8, 0x00a7c8},
+ {0x00a7ca, 0x00a7ca},
+ {0x00a7d1, 0x00a7d1},
+ {0x00a7d3, 0x00a7d3},
+ {0x00a7d5, 0x00a7d5},
+ {0x00a7d7, 0x00a7d7},
+ {0x00a7d9, 0x00a7d9},
+ {0x00a7f2, 0x00a7f4},
+ {0x00a7f6, 0x00a7f6},
+ {0x00a7f8, 0x00a7f9},
+ {0x00a7fa, 0x00a7fa},
+ {0x00ab30, 0x00ab5a},
+ {0x00ab5c, 0x00ab5f},
+ {0x00ab60, 0x00ab68},
+ {0x00ab69, 0x00ab69},
+ {0x00ab70, 0x00abbf},
+ {0x00fb00, 0x00fb06},
+ {0x00fb13, 0x00fb17},
+ {0x00ff41, 0x00ff5a},
+ {0x010428, 0x01044f},
+ {0x0104d8, 0x0104fb},
+ {0x010597, 0x0105a1},
+ {0x0105a3, 0x0105b1},
+ {0x0105b3, 0x0105b9},
+ {0x0105bb, 0x0105bc},
+ {0x010780, 0x010780},
+ {0x010783, 0x010785},
+ {0x010787, 0x0107b0},
+ {0x0107b2, 0x0107ba},
+ {0x010cc0, 0x010cf2},
+ {0x0118c0, 0x0118df},
+ {0x016e60, 0x016e7f},
+ {0x01d41a, 0x01d433},
+ {0x01d44e, 0x01d454},
+ {0x01d456, 0x01d467},
+ {0x01d482, 0x01d49b},
+ {0x01d4b6, 0x01d4b9},
+ {0x01d4bb, 0x01d4bb},
+ {0x01d4bd, 0x01d4c3},
+ {0x01d4c5, 0x01d4cf},
+ {0x01d4ea, 0x01d503},
+ {0x01d51e, 0x01d537},
+ {0x01d552, 0x01d56b},
+ {0x01d586, 0x01d59f},
+ {0x01d5ba, 0x01d5d3},
+ {0x01d5ee, 0x01d607},
+ {0x01d622, 0x01d63b},
+ {0x01d656, 0x01d66f},
+ {0x01d68a, 0x01d6a5},
+ {0x01d6c2, 0x01d6da},
+ {0x01d6dc, 0x01d6e1},
+ {0x01d6fc, 0x01d714},
+ {0x01d716, 0x01d71b},
+ {0x01d736, 0x01d74e},
+ {0x01d750, 0x01d755},
+ {0x01d770, 0x01d788},
+ {0x01d78a, 0x01d78f},
+ {0x01d7aa, 0x01d7c2},
+ {0x01d7c4, 0x01d7c9},
+ {0x01d7cb, 0x01d7cb},
+ {0x01df00, 0x01df09},
+ {0x01df0b, 0x01df1e},
+ {0x01df25, 0x01df2a},
+ {0x01e030, 0x01e06d},
+ {0x01e922, 0x01e943}
+};
+
+/* table of Unicode codepoint ranges of Uppercase characters */
+static const pg_unicode_range unicode_uppercase[651] =
+{
+ {0x000041, 0x00005a},
+ {0x0000c0, 0x0000d6},
+ {0x0000d8, 0x0000de},
+ {0x000100, 0x000100},
+ {0x000102, 0x000102},
+ {0x000104, 0x000104},
+ {0x000106, 0x000106},
+ {0x000108, 0x000108},
+ {0x00010a, 0x00010a},
+ {0x00010c, 0x00010c},
+ {0x00010e, 0x00010e},
+ {0x000110, 0x000110},
+ {0x000112, 0x000112},
+ {0x000114, 0x000114},
+ {0x000116, 0x000116},
+ {0x000118, 0x000118},
+ {0x00011a, 0x00011a},
+ {0x00011c, 0x00011c},
+ {0x00011e, 0x00011e},
+ {0x000120, 0x000120},
+ {0x000122, 0x000122},
+ {0x000124, 0x000124},
+ {0x000126, 0x000126},
+ {0x000128, 0x000128},
+ {0x00012a, 0x00012a},
+ {0x00012c, 0x00012c},
+ {0x00012e, 0x00012e},
+ {0x000130, 0x000130},
+ {0x000132, 0x000132},
+ {0x000134, 0x000134},
+ {0x000136, 0x000136},
+ {0x000139, 0x000139},
+ {0x00013b, 0x00013b},
+ {0x00013d, 0x00013d},
+ {0x00013f, 0x00013f},
+ {0x000141, 0x000141},
+ {0x000143, 0x000143},
+ {0x000145, 0x000145},
+ {0x000147, 0x000147},
+ {0x00014a, 0x00014a},
+ {0x00014c, 0x00014c},
+ {0x00014e, 0x00014e},
+ {0x000150, 0x000150},
+ {0x000152, 0x000152},
+ {0x000154, 0x000154},
+ {0x000156, 0x000156},
+ {0x000158, 0x000158},
+ {0x00015a, 0x00015a},
+ {0x00015c, 0x00015c},
+ {0x00015e, 0x00015e},
+ {0x000160, 0x000160},
+ {0x000162, 0x000162},
+ {0x000164, 0x000164},
+ {0x000166, 0x000166},
+ {0x000168, 0x000168},
+ {0x00016a, 0x00016a},
+ {0x00016c, 0x00016c},
+ {0x00016e, 0x00016e},
+ {0x000170, 0x000170},
+ {0x000172, 0x000172},
+ {0x000174, 0x000174},
+ {0x000176, 0x000176},
+ {0x000178, 0x000179},
+ {0x00017b, 0x00017b},
+ {0x00017d, 0x00017d},
+ {0x000181, 0x000182},
+ {0x000184, 0x000184},
+ {0x000186, 0x000187},
+ {0x000189, 0x00018b},
+ {0x00018e, 0x000191},
+ {0x000193, 0x000194},
+ {0x000196, 0x000198},
+ {0x00019c, 0x00019d},
+ {0x00019f, 0x0001a0},
+ {0x0001a2, 0x0001a2},
+ {0x0001a4, 0x0001a4},
+ {0x0001a6, 0x0001a7},
+ {0x0001a9, 0x0001a9},
+ {0x0001ac, 0x0001ac},
+ {0x0001ae, 0x0001af},
+ {0x0001b1, 0x0001b3},
+ {0x0001b5, 0x0001b5},
+ {0x0001b7, 0x0001b8},
+ {0x0001bc, 0x0001bc},
+ {0x0001c4, 0x0001c4},
+ {0x0001c7, 0x0001c7},
+ {0x0001ca, 0x0001ca},
+ {0x0001cd, 0x0001cd},
+ {0x0001cf, 0x0001cf},
+ {0x0001d1, 0x0001d1},
+ {0x0001d3, 0x0001d3},
+ {0x0001d5, 0x0001d5},
+ {0x0001d7, 0x0001d7},
+ {0x0001d9, 0x0001d9},
+ {0x0001db, 0x0001db},
+ {0x0001de, 0x0001de},
+ {0x0001e0, 0x0001e0},
+ {0x0001e2, 0x0001e2},
+ {0x0001e4, 0x0001e4},
+ {0x0001e6, 0x0001e6},
+ {0x0001e8, 0x0001e8},
+ {0x0001ea, 0x0001ea},
+ {0x0001ec, 0x0001ec},
+ {0x0001ee, 0x0001ee},
+ {0x0001f1, 0x0001f1},
+ {0x0001f4, 0x0001f4},
+ {0x0001f6, 0x0001f8},
+ {0x0001fa, 0x0001fa},
+ {0x0001fc, 0x0001fc},
+ {0x0001fe, 0x0001fe},
+ {0x000200, 0x000200},
+ {0x000202, 0x000202},
+ {0x000204, 0x000204},
+ {0x000206, 0x000206},
+ {0x000208, 0x000208},
+ {0x00020a, 0x00020a},
+ {0x00020c, 0x00020c},
+ {0x00020e, 0x00020e},
+ {0x000210, 0x000210},
+ {0x000212, 0x000212},
+ {0x000214, 0x000214},
+ {0x000216, 0x000216},
+ {0x000218, 0x000218},
+ {0x00021a, 0x00021a},
+ {0x00021c, 0x00021c},
+ {0x00021e, 0x00021e},
+ {0x000220, 0x000220},
+ {0x000222, 0x000222},
+ {0x000224, 0x000224},
+ {0x000226, 0x000226},
+ {0x000228, 0x000228},
+ {0x00022a, 0x00022a},
+ {0x00022c, 0x00022c},
+ {0x00022e, 0x00022e},
+ {0x000230, 0x000230},
+ {0x000232, 0x000232},
+ {0x00023a, 0x00023b},
+ {0x00023d, 0x00023e},
+ {0x000241, 0x000241},
+ {0x000243, 0x000246},
+ {0x000248, 0x000248},
+ {0x00024a, 0x00024a},
+ {0x00024c, 0x00024c},
+ {0x00024e, 0x00024e},
+ {0x000370, 0x000370},
+ {0x000372, 0x000372},
+ {0x000376, 0x000376},
+ {0x00037f, 0x00037f},
+ {0x000386, 0x000386},
+ {0x000388, 0x00038a},
+ {0x00038c, 0x00038c},
+ {0x00038e, 0x00038f},
+ {0x000391, 0x0003a1},
+ {0x0003a3, 0x0003ab},
+ {0x0003cf, 0x0003cf},
+ {0x0003d2, 0x0003d4},
+ {0x0003d8, 0x0003d8},
+ {0x0003da, 0x0003da},
+ {0x0003dc, 0x0003dc},
+ {0x0003de, 0x0003de},
+ {0x0003e0, 0x0003e0},
+ {0x0003e2, 0x0003e2},
+ {0x0003e4, 0x0003e4},
+ {0x0003e6, 0x0003e6},
+ {0x0003e8, 0x0003e8},
+ {0x0003ea, 0x0003ea},
+ {0x0003ec, 0x0003ec},
+ {0x0003ee, 0x0003ee},
+ {0x0003f4, 0x0003f4},
+ {0x0003f7, 0x0003f7},
+ {0x0003f9, 0x0003fa},
+ {0x0003fd, 0x00042f},
+ {0x000460, 0x000460},
+ {0x000462, 0x000462},
+ {0x000464, 0x000464},
+ {0x000466, 0x000466},
+ {0x000468, 0x000468},
+ {0x00046a, 0x00046a},
+ {0x00046c, 0x00046c},
+ {0x00046e, 0x00046e},
+ {0x000470, 0x000470},
+ {0x000472, 0x000472},
+ {0x000474, 0x000474},
+ {0x000476, 0x000476},
+ {0x000478, 0x000478},
+ {0x00047a, 0x00047a},
+ {0x00047c, 0x00047c},
+ {0x00047e, 0x00047e},
+ {0x000480, 0x000480},
+ {0x00048a, 0x00048a},
+ {0x00048c, 0x00048c},
+ {0x00048e, 0x00048e},
+ {0x000490, 0x000490},
+ {0x000492, 0x000492},
+ {0x000494, 0x000494},
+ {0x000496, 0x000496},
+ {0x000498, 0x000498},
+ {0x00049a, 0x00049a},
+ {0x00049c, 0x00049c},
+ {0x00049e, 0x00049e},
+ {0x0004a0, 0x0004a0},
+ {0x0004a2, 0x0004a2},
+ {0x0004a4, 0x0004a4},
+ {0x0004a6, 0x0004a6},
+ {0x0004a8, 0x0004a8},
+ {0x0004aa, 0x0004aa},
+ {0x0004ac, 0x0004ac},
+ {0x0004ae, 0x0004ae},
+ {0x0004b0, 0x0004b0},
+ {0x0004b2, 0x0004b2},
+ {0x0004b4, 0x0004b4},
+ {0x0004b6, 0x0004b6},
+ {0x0004b8, 0x0004b8},
+ {0x0004ba, 0x0004ba},
+ {0x0004bc, 0x0004bc},
+ {0x0004be, 0x0004be},
+ {0x0004c0, 0x0004c1},
+ {0x0004c3, 0x0004c3},
+ {0x0004c5, 0x0004c5},
+ {0x0004c7, 0x0004c7},
+ {0x0004c9, 0x0004c9},
+ {0x0004cb, 0x0004cb},
+ {0x0004cd, 0x0004cd},
+ {0x0004d0, 0x0004d0},
+ {0x0004d2, 0x0004d2},
+ {0x0004d4, 0x0004d4},
+ {0x0004d6, 0x0004d6},
+ {0x0004d8, 0x0004d8},
+ {0x0004da, 0x0004da},
+ {0x0004dc, 0x0004dc},
+ {0x0004de, 0x0004de},
+ {0x0004e0, 0x0004e0},
+ {0x0004e2, 0x0004e2},
+ {0x0004e4, 0x0004e4},
+ {0x0004e6, 0x0004e6},
+ {0x0004e8, 0x0004e8},
+ {0x0004ea, 0x0004ea},
+ {0x0004ec, 0x0004ec},
+ {0x0004ee, 0x0004ee},
+ {0x0004f0, 0x0004f0},
+ {0x0004f2, 0x0004f2},
+ {0x0004f4, 0x0004f4},
+ {0x0004f6, 0x0004f6},
+ {0x0004f8, 0x0004f8},
+ {0x0004fa, 0x0004fa},
+ {0x0004fc, 0x0004fc},
+ {0x0004fe, 0x0004fe},
+ {0x000500, 0x000500},
+ {0x000502, 0x000502},
+ {0x000504, 0x000504},
+ {0x000506, 0x000506},
+ {0x000508, 0x000508},
+ {0x00050a, 0x00050a},
+ {0x00050c, 0x00050c},
+ {0x00050e, 0x00050e},
+ {0x000510, 0x000510},
+ {0x000512, 0x000512},
+ {0x000514, 0x000514},
+ {0x000516, 0x000516},
+ {0x000518, 0x000518},
+ {0x00051a, 0x00051a},
+ {0x00051c, 0x00051c},
+ {0x00051e, 0x00051e},
+ {0x000520, 0x000520},
+ {0x000522, 0x000522},
+ {0x000524, 0x000524},
+ {0x000526, 0x000526},
+ {0x000528, 0x000528},
+ {0x00052a, 0x00052a},
+ {0x00052c, 0x00052c},
+ {0x00052e, 0x00052e},
+ {0x000531, 0x000556},
+ {0x0010a0, 0x0010c5},
+ {0x0010c7, 0x0010c7},
+ {0x0010cd, 0x0010cd},
+ {0x0013a0, 0x0013f5},
+ {0x001c90, 0x001cba},
+ {0x001cbd, 0x001cbf},
+ {0x001e00, 0x001e00},
+ {0x001e02, 0x001e02},
+ {0x001e04, 0x001e04},
+ {0x001e06, 0x001e06},
+ {0x001e08, 0x001e08},
+ {0x001e0a, 0x001e0a},
+ {0x001e0c, 0x001e0c},
+ {0x001e0e, 0x001e0e},
+ {0x001e10, 0x001e10},
+ {0x001e12, 0x001e12},
+ {0x001e14, 0x001e14},
+ {0x001e16, 0x001e16},
+ {0x001e18, 0x001e18},
+ {0x001e1a, 0x001e1a},
+ {0x001e1c, 0x001e1c},
+ {0x001e1e, 0x001e1e},
+ {0x001e20, 0x001e20},
+ {0x001e22, 0x001e22},
+ {0x001e24, 0x001e24},
+ {0x001e26, 0x001e26},
+ {0x001e28, 0x001e28},
+ {0x001e2a, 0x001e2a},
+ {0x001e2c, 0x001e2c},
+ {0x001e2e, 0x001e2e},
+ {0x001e30, 0x001e30},
+ {0x001e32, 0x001e32},
+ {0x001e34, 0x001e34},
+ {0x001e36, 0x001e36},
+ {0x001e38, 0x001e38},
+ {0x001e3a, 0x001e3a},
+ {0x001e3c, 0x001e3c},
+ {0x001e3e, 0x001e3e},
+ {0x001e40, 0x001e40},
+ {0x001e42, 0x001e42},
+ {0x001e44, 0x001e44},
+ {0x001e46, 0x001e46},
+ {0x001e48, 0x001e48},
+ {0x001e4a, 0x001e4a},
+ {0x001e4c, 0x001e4c},
+ {0x001e4e, 0x001e4e},
+ {0x001e50, 0x001e50},
+ {0x001e52, 0x001e52},
+ {0x001e54, 0x001e54},
+ {0x001e56, 0x001e56},
+ {0x001e58, 0x001e58},
+ {0x001e5a, 0x001e5a},
+ {0x001e5c, 0x001e5c},
+ {0x001e5e, 0x001e5e},
+ {0x001e60, 0x001e60},
+ {0x001e62, 0x001e62},
+ {0x001e64, 0x001e64},
+ {0x001e66, 0x001e66},
+ {0x001e68, 0x001e68},
+ {0x001e6a, 0x001e6a},
+ {0x001e6c, 0x001e6c},
+ {0x001e6e, 0x001e6e},
+ {0x001e70, 0x001e70},
+ {0x001e72, 0x001e72},
+ {0x001e74, 0x001e74},
+ {0x001e76, 0x001e76},
+ {0x001e78, 0x001e78},
+ {0x001e7a, 0x001e7a},
+ {0x001e7c, 0x001e7c},
+ {0x001e7e, 0x001e7e},
+ {0x001e80, 0x001e80},
+ {0x001e82, 0x001e82},
+ {0x001e84, 0x001e84},
+ {0x001e86, 0x001e86},
+ {0x001e88, 0x001e88},
+ {0x001e8a, 0x001e8a},
+ {0x001e8c, 0x001e8c},
+ {0x001e8e, 0x001e8e},
+ {0x001e90, 0x001e90},
+ {0x001e92, 0x001e92},
+ {0x001e94, 0x001e94},
+ {0x001e9e, 0x001e9e},
+ {0x001ea0, 0x001ea0},
+ {0x001ea2, 0x001ea2},
+ {0x001ea4, 0x001ea4},
+ {0x001ea6, 0x001ea6},
+ {0x001ea8, 0x001ea8},
+ {0x001eaa, 0x001eaa},
+ {0x001eac, 0x001eac},
+ {0x001eae, 0x001eae},
+ {0x001eb0, 0x001eb0},
+ {0x001eb2, 0x001eb2},
+ {0x001eb4, 0x001eb4},
+ {0x001eb6, 0x001eb6},
+ {0x001eb8, 0x001eb8},
+ {0x001eba, 0x001eba},
+ {0x001ebc, 0x001ebc},
+ {0x001ebe, 0x001ebe},
+ {0x001ec0, 0x001ec0},
+ {0x001ec2, 0x001ec2},
+ {0x001ec4, 0x001ec4},
+ {0x001ec6, 0x001ec6},
+ {0x001ec8, 0x001ec8},
+ {0x001eca, 0x001eca},
+ {0x001ecc, 0x001ecc},
+ {0x001ece, 0x001ece},
+ {0x001ed0, 0x001ed0},
+ {0x001ed2, 0x001ed2},
+ {0x001ed4, 0x001ed4},
+ {0x001ed6, 0x001ed6},
+ {0x001ed8, 0x001ed8},
+ {0x001eda, 0x001eda},
+ {0x001edc, 0x001edc},
+ {0x001ede, 0x001ede},
+ {0x001ee0, 0x001ee0},
+ {0x001ee2, 0x001ee2},
+ {0x001ee4, 0x001ee4},
+ {0x001ee6, 0x001ee6},
+ {0x001ee8, 0x001ee8},
+ {0x001eea, 0x001eea},
+ {0x001eec, 0x001eec},
+ {0x001eee, 0x001eee},
+ {0x001ef0, 0x001ef0},
+ {0x001ef2, 0x001ef2},
+ {0x001ef4, 0x001ef4},
+ {0x001ef6, 0x001ef6},
+ {0x001ef8, 0x001ef8},
+ {0x001efa, 0x001efa},
+ {0x001efc, 0x001efc},
+ {0x001efe, 0x001efe},
+ {0x001f08, 0x001f0f},
+ {0x001f18, 0x001f1d},
+ {0x001f28, 0x001f2f},
+ {0x001f38, 0x001f3f},
+ {0x001f48, 0x001f4d},
+ {0x001f59, 0x001f59},
+ {0x001f5b, 0x001f5b},
+ {0x001f5d, 0x001f5d},
+ {0x001f5f, 0x001f5f},
+ {0x001f68, 0x001f6f},
+ {0x001fb8, 0x001fbb},
+ {0x001fc8, 0x001fcb},
+ {0x001fd8, 0x001fdb},
+ {0x001fe8, 0x001fec},
+ {0x001ff8, 0x001ffb},
+ {0x002102, 0x002102},
+ {0x002107, 0x002107},
+ {0x00210b, 0x00210d},
+ {0x002110, 0x002112},
+ {0x002115, 0x002115},
+ {0x002119, 0x00211d},
+ {0x002124, 0x002124},
+ {0x002126, 0x002126},
+ {0x002128, 0x002128},
+ {0x00212a, 0x00212d},
+ {0x002130, 0x002133},
+ {0x00213e, 0x00213f},
+ {0x002145, 0x002145},
+ {0x002160, 0x00216f},
+ {0x002183, 0x002183},
+ {0x0024b6, 0x0024cf},
+ {0x002c00, 0x002c2f},
+ {0x002c60, 0x002c60},
+ {0x002c62, 0x002c64},
+ {0x002c67, 0x002c67},
+ {0x002c69, 0x002c69},
+ {0x002c6b, 0x002c6b},
+ {0x002c6d, 0x002c70},
+ {0x002c72, 0x002c72},
+ {0x002c75, 0x002c75},
+ {0x002c7e, 0x002c80},
+ {0x002c82, 0x002c82},
+ {0x002c84, 0x002c84},
+ {0x002c86, 0x002c86},
+ {0x002c88, 0x002c88},
+ {0x002c8a, 0x002c8a},
+ {0x002c8c, 0x002c8c},
+ {0x002c8e, 0x002c8e},
+ {0x002c90, 0x002c90},
+ {0x002c92, 0x002c92},
+ {0x002c94, 0x002c94},
+ {0x002c96, 0x002c96},
+ {0x002c98, 0x002c98},
+ {0x002c9a, 0x002c9a},
+ {0x002c9c, 0x002c9c},
+ {0x002c9e, 0x002c9e},
+ {0x002ca0, 0x002ca0},
+ {0x002ca2, 0x002ca2},
+ {0x002ca4, 0x002ca4},
+ {0x002ca6, 0x002ca6},
+ {0x002ca8, 0x002ca8},
+ {0x002caa, 0x002caa},
+ {0x002cac, 0x002cac},
+ {0x002cae, 0x002cae},
+ {0x002cb0, 0x002cb0},
+ {0x002cb2, 0x002cb2},
+ {0x002cb4, 0x002cb4},
+ {0x002cb6, 0x002cb6},
+ {0x002cb8, 0x002cb8},
+ {0x002cba, 0x002cba},
+ {0x002cbc, 0x002cbc},
+ {0x002cbe, 0x002cbe},
+ {0x002cc0, 0x002cc0},
+ {0x002cc2, 0x002cc2},
+ {0x002cc4, 0x002cc4},
+ {0x002cc6, 0x002cc6},
+ {0x002cc8, 0x002cc8},
+ {0x002cca, 0x002cca},
+ {0x002ccc, 0x002ccc},
+ {0x002cce, 0x002cce},
+ {0x002cd0, 0x002cd0},
+ {0x002cd2, 0x002cd2},
+ {0x002cd4, 0x002cd4},
+ {0x002cd6, 0x002cd6},
+ {0x002cd8, 0x002cd8},
+ {0x002cda, 0x002cda},
+ {0x002cdc, 0x002cdc},
+ {0x002cde, 0x002cde},
+ {0x002ce0, 0x002ce0},
+ {0x002ce2, 0x002ce2},
+ {0x002ceb, 0x002ceb},
+ {0x002ced, 0x002ced},
+ {0x002cf2, 0x002cf2},
+ {0x00a640, 0x00a640},
+ {0x00a642, 0x00a642},
+ {0x00a644, 0x00a644},
+ {0x00a646, 0x00a646},
+ {0x00a648, 0x00a648},
+ {0x00a64a, 0x00a64a},
+ {0x00a64c, 0x00a64c},
+ {0x00a64e, 0x00a64e},
+ {0x00a650, 0x00a650},
+ {0x00a652, 0x00a652},
+ {0x00a654, 0x00a654},
+ {0x00a656, 0x00a656},
+ {0x00a658, 0x00a658},
+ {0x00a65a, 0x00a65a},
+ {0x00a65c, 0x00a65c},
+ {0x00a65e, 0x00a65e},
+ {0x00a660, 0x00a660},
+ {0x00a662, 0x00a662},
+ {0x00a664, 0x00a664},
+ {0x00a666, 0x00a666},
+ {0x00a668, 0x00a668},
+ {0x00a66a, 0x00a66a},
+ {0x00a66c, 0x00a66c},
+ {0x00a680, 0x00a680},
+ {0x00a682, 0x00a682},
+ {0x00a684, 0x00a684},
+ {0x00a686, 0x00a686},
+ {0x00a688, 0x00a688},
+ {0x00a68a, 0x00a68a},
+ {0x00a68c, 0x00a68c},
+ {0x00a68e, 0x00a68e},
+ {0x00a690, 0x00a690},
+ {0x00a692, 0x00a692},
+ {0x00a694, 0x00a694},
+ {0x00a696, 0x00a696},
+ {0x00a698, 0x00a698},
+ {0x00a69a, 0x00a69a},
+ {0x00a722, 0x00a722},
+ {0x00a724, 0x00a724},
+ {0x00a726, 0x00a726},
+ {0x00a728, 0x00a728},
+ {0x00a72a, 0x00a72a},
+ {0x00a72c, 0x00a72c},
+ {0x00a72e, 0x00a72e},
+ {0x00a732, 0x00a732},
+ {0x00a734, 0x00a734},
+ {0x00a736, 0x00a736},
+ {0x00a738, 0x00a738},
+ {0x00a73a, 0x00a73a},
+ {0x00a73c, 0x00a73c},
+ {0x00a73e, 0x00a73e},
+ {0x00a740, 0x00a740},
+ {0x00a742, 0x00a742},
+ {0x00a744, 0x00a744},
+ {0x00a746, 0x00a746},
+ {0x00a748, 0x00a748},
+ {0x00a74a, 0x00a74a},
+ {0x00a74c, 0x00a74c},
+ {0x00a74e, 0x00a74e},
+ {0x00a750, 0x00a750},
+ {0x00a752, 0x00a752},
+ {0x00a754, 0x00a754},
+ {0x00a756, 0x00a756},
+ {0x00a758, 0x00a758},
+ {0x00a75a, 0x00a75a},
+ {0x00a75c, 0x00a75c},
+ {0x00a75e, 0x00a75e},
+ {0x00a760, 0x00a760},
+ {0x00a762, 0x00a762},
+ {0x00a764, 0x00a764},
+ {0x00a766, 0x00a766},
+ {0x00a768, 0x00a768},
+ {0x00a76a, 0x00a76a},
+ {0x00a76c, 0x00a76c},
+ {0x00a76e, 0x00a76e},
+ {0x00a779, 0x00a779},
+ {0x00a77b, 0x00a77b},
+ {0x00a77d, 0x00a77e},
+ {0x00a780, 0x00a780},
+ {0x00a782, 0x00a782},
+ {0x00a784, 0x00a784},
+ {0x00a786, 0x00a786},
+ {0x00a78b, 0x00a78b},
+ {0x00a78d, 0x00a78d},
+ {0x00a790, 0x00a790},
+ {0x00a792, 0x00a792},
+ {0x00a796, 0x00a796},
+ {0x00a798, 0x00a798},
+ {0x00a79a, 0x00a79a},
+ {0x00a79c, 0x00a79c},
+ {0x00a79e, 0x00a79e},
+ {0x00a7a0, 0x00a7a0},
+ {0x00a7a2, 0x00a7a2},
+ {0x00a7a4, 0x00a7a4},
+ {0x00a7a6, 0x00a7a6},
+ {0x00a7a8, 0x00a7a8},
+ {0x00a7aa, 0x00a7ae},
+ {0x00a7b0, 0x00a7b4},
+ {0x00a7b6, 0x00a7b6},
+ {0x00a7b8, 0x00a7b8},
+ {0x00a7ba, 0x00a7ba},
+ {0x00a7bc, 0x00a7bc},
+ {0x00a7be, 0x00a7be},
+ {0x00a7c0, 0x00a7c0},
+ {0x00a7c2, 0x00a7c2},
+ {0x00a7c4, 0x00a7c7},
+ {0x00a7c9, 0x00a7c9},
+ {0x00a7d0, 0x00a7d0},
+ {0x00a7d6, 0x00a7d6},
+ {0x00a7d8, 0x00a7d8},
+ {0x00a7f5, 0x00a7f5},
+ {0x00ff21, 0x00ff3a},
+ {0x010400, 0x010427},
+ {0x0104b0, 0x0104d3},
+ {0x010570, 0x01057a},
+ {0x01057c, 0x01058a},
+ {0x01058c, 0x010592},
+ {0x010594, 0x010595},
+ {0x010c80, 0x010cb2},
+ {0x0118a0, 0x0118bf},
+ {0x016e40, 0x016e5f},
+ {0x01d400, 0x01d419},
+ {0x01d434, 0x01d44d},
+ {0x01d468, 0x01d481},
+ {0x01d49c, 0x01d49c},
+ {0x01d49e, 0x01d49f},
+ {0x01d4a2, 0x01d4a2},
+ {0x01d4a5, 0x01d4a6},
+ {0x01d4a9, 0x01d4ac},
+ {0x01d4ae, 0x01d4b5},
+ {0x01d4d0, 0x01d4e9},
+ {0x01d504, 0x01d505},
+ {0x01d507, 0x01d50a},
+ {0x01d50d, 0x01d514},
+ {0x01d516, 0x01d51c},
+ {0x01d538, 0x01d539},
+ {0x01d53b, 0x01d53e},
+ {0x01d540, 0x01d544},
+ {0x01d546, 0x01d546},
+ {0x01d54a, 0x01d550},
+ {0x01d56c, 0x01d585},
+ {0x01d5a0, 0x01d5b9},
+ {0x01d5d4, 0x01d5ed},
+ {0x01d608, 0x01d621},
+ {0x01d63c, 0x01d655},
+ {0x01d670, 0x01d689},
+ {0x01d6a8, 0x01d6c0},
+ {0x01d6e2, 0x01d6fa},
+ {0x01d71c, 0x01d734},
+ {0x01d756, 0x01d76e},
+ {0x01d790, 0x01d7a8},
+ {0x01d7ca, 0x01d7ca},
+ {0x01e900, 0x01e921},
+ {0x01f130, 0x01f149},
+ {0x01f150, 0x01f169},
+ {0x01f170, 0x01f189}
+};
+
+/* table of Unicode codepoint ranges of White_Space characters */
+static const pg_unicode_range unicode_white_space[11] =
+{
+ {0x000009, 0x00000d},
+ {0x000020, 0x000020},
+ {0x000085, 0x000085},
+ {0x0000a0, 0x0000a0},
+ {0x001680, 0x001680},
+ {0x002000, 0x00200a},
+ {0x002028, 0x002028},
+ {0x002029, 0x002029},
+ {0x00202f, 0x00202f},
+ {0x00205f, 0x00205f},
+ {0x003000, 0x003000}
+};
+
+/* table of Unicode codepoint ranges of Hex_Digit characters */
+static const pg_unicode_range unicode_hex_digit[6] =
+{
+ {0x000030, 0x000039},
+ {0x000041, 0x000046},
+ {0x000061, 0x000066},
+ {0x00ff10, 0x00ff19},
+ {0x00ff21, 0x00ff26},
+ {0x00ff41, 0x00ff46}
+};
+
+/* table of Unicode codepoint ranges of Join_Control characters */
+static const pg_unicode_range unicode_join_control[1] =
+{
+ {0x00200c, 0x00200d}
+};
--
2.34.1
[text/x-patch] v2-0002-Shrink-unicode-category-table.patch (101.7K, ../../[email protected]/5-v2-0002-Shrink-unicode-category-table.patch)
download | inline diff:
From 35cd57cd65205573be3a3eff91affe307da405d0 Mon Sep 17 00:00:00 2001
From: Jeff Davis <[email protected]>
Date: Wed, 22 Nov 2023 11:30:31 -0800
Subject: [PATCH v2 2/3] Shrink unicode category table.
Missing entries can implicitly be considered "unassigned".
---
.../generate-unicode_category_table.pl | 21 +-
src/common/unicode_category.c | 6 +-
src/include/common/unicode_category_table.h | 711 +-----------------
3 files changed, 15 insertions(+), 723 deletions(-)
diff --git a/src/common/unicode/generate-unicode_category_table.pl b/src/common/unicode/generate-unicode_category_table.pl
index 8f03425e0b..992b877ede 100644
--- a/src/common/unicode/generate-unicode_category_table.pl
+++ b/src/common/unicode/generate-unicode_category_table.pl
@@ -72,7 +72,10 @@ while (my $line = <$FH>)
# the current range, emit the current range and initialize a new
# range representing the gap.
if ($range_end + 1 != $code && $range_category ne $gap_category) {
- push(@category_ranges, {start => $range_start, end => $range_end, category => $range_category});
+ if ($range_category ne $CATEGORY_UNASSIGNED) {
+ push(@category_ranges, {start => $range_start, end => $range_end,
+ category => $range_category});
+ }
$range_start = $range_end + 1;
$range_end = $code - 1;
$range_category = $gap_category;
@@ -80,7 +83,10 @@ while (my $line = <$FH>)
# different category; new range
if ($range_category ne $category) {
- push(@category_ranges, {start => $range_start, end => $range_end, category => $range_category});
+ if ($range_category ne $CATEGORY_UNASSIGNED) {
+ push(@category_ranges, {start => $range_start, end => $range_end,
+ category => $range_category});
+ }
$range_start = $code;
$range_end = $code;
$range_category = $category;
@@ -109,14 +115,9 @@ die "<..., First> entry with no corresponding <..., Last> entry"
if $gap_category ne $CATEGORY_UNASSIGNED;
# emit final range
-push(@category_ranges, {start => $range_start, end => $range_end, category => $range_category});
-
-# emit range for any unassigned code points after last entry
-if ($range_end < 0x10FFFF) {
- $range_start = $range_end + 1;
- $range_end = 0x10FFFF;
- $range_category = $CATEGORY_UNASSIGNED;
- push(@category_ranges, {start => $range_start, end => $range_end, category => $range_category});
+if ($range_category ne $CATEGORY_UNASSIGNED) {
+ push(@category_ranges, {start => $range_start, end => $range_end,
+ category => $range_category});
}
my $num_ranges = scalar @category_ranges;
diff --git a/src/common/unicode_category.c b/src/common/unicode_category.c
index cec9c0d998..189cd6eca3 100644
--- a/src/common/unicode_category.c
+++ b/src/common/unicode_category.c
@@ -28,8 +28,7 @@ unicode_category(pg_wchar ucs)
int mid;
int max = lengthof(unicode_categories) - 1;
- Assert(ucs >= unicode_categories[0].first &&
- ucs <= unicode_categories[max].last);
+ Assert(ucs <= 0x10ffff);
while (max >= min)
{
@@ -42,8 +41,7 @@ unicode_category(pg_wchar ucs)
return unicode_categories[mid].category;
}
- Assert(false);
- return (pg_unicode_category) - 1;
+ return PG_U_UNASSIGNED;
}
/*
diff --git a/src/include/common/unicode_category_table.h b/src/include/common/unicode_category_table.h
index 06ad50d215..14f1ea0677 100644
--- a/src/include/common/unicode_category_table.h
+++ b/src/include/common/unicode_category_table.h
@@ -26,7 +26,7 @@ typedef struct
} pg_category_range;
/* table of Unicode codepoint ranges and their categories */
-static const pg_category_range unicode_categories[4009] =
+static const pg_category_range unicode_categories[3302] =
{
{0x000000, 0x00001f, PG_U_CONTROL},
{0x000020, 0x000020, PG_U_SPACE_SEPARATOR},
@@ -397,23 +397,18 @@ static const pg_category_range unicode_categories[4009] =
{0x000375, 0x000375, PG_U_MODIFIER_SYMBOL},
{0x000376, 0x000376, PG_U_UPPERCASE_LETTER},
{0x000377, 0x000377, PG_U_LOWERCASE_LETTER},
- {0x000378, 0x000379, PG_U_UNASSIGNED},
{0x00037a, 0x00037a, PG_U_MODIFIER_LETTER},
{0x00037b, 0x00037d, PG_U_LOWERCASE_LETTER},
{0x00037e, 0x00037e, PG_U_OTHER_PUNCTUATION},
{0x00037f, 0x00037f, PG_U_UPPERCASE_LETTER},
- {0x000380, 0x000383, PG_U_UNASSIGNED},
{0x000384, 0x000385, PG_U_MODIFIER_SYMBOL},
{0x000386, 0x000386, PG_U_UPPERCASE_LETTER},
{0x000387, 0x000387, PG_U_OTHER_PUNCTUATION},
{0x000388, 0x00038a, PG_U_UPPERCASE_LETTER},
- {0x00038b, 0x00038b, PG_U_UNASSIGNED},
{0x00038c, 0x00038c, PG_U_UPPERCASE_LETTER},
- {0x00038d, 0x00038d, PG_U_UNASSIGNED},
{0x00038e, 0x00038f, PG_U_UPPERCASE_LETTER},
{0x000390, 0x000390, PG_U_LOWERCASE_LETTER},
{0x000391, 0x0003a1, PG_U_UPPERCASE_LETTER},
- {0x0003a2, 0x0003a2, PG_U_UNASSIGNED},
{0x0003a3, 0x0003ab, PG_U_UPPERCASE_LETTER},
{0x0003ac, 0x0003ce, PG_U_LOWERCASE_LETTER},
{0x0003cf, 0x0003cf, PG_U_UPPERCASE_LETTER},
@@ -654,18 +649,14 @@ static const pg_category_range unicode_categories[4009] =
{0x00052d, 0x00052d, PG_U_LOWERCASE_LETTER},
{0x00052e, 0x00052e, PG_U_UPPERCASE_LETTER},
{0x00052f, 0x00052f, PG_U_LOWERCASE_LETTER},
- {0x000530, 0x000530, PG_U_UNASSIGNED},
{0x000531, 0x000556, PG_U_UPPERCASE_LETTER},
- {0x000557, 0x000558, PG_U_UNASSIGNED},
{0x000559, 0x000559, PG_U_MODIFIER_LETTER},
{0x00055a, 0x00055f, PG_U_OTHER_PUNCTUATION},
{0x000560, 0x000588, PG_U_LOWERCASE_LETTER},
{0x000589, 0x000589, PG_U_OTHER_PUNCTUATION},
{0x00058a, 0x00058a, PG_U_DASH_PUNCTUATION},
- {0x00058b, 0x00058c, PG_U_UNASSIGNED},
{0x00058d, 0x00058e, PG_U_OTHER_SYMBOL},
{0x00058f, 0x00058f, PG_U_CURRENCY_SYMBOL},
- {0x000590, 0x000590, PG_U_UNASSIGNED},
{0x000591, 0x0005bd, PG_U_NONSPACING_MARK},
{0x0005be, 0x0005be, PG_U_DASH_PUNCTUATION},
{0x0005bf, 0x0005bf, PG_U_NONSPACING_MARK},
@@ -675,12 +666,9 @@ static const pg_category_range unicode_categories[4009] =
{0x0005c4, 0x0005c5, PG_U_NONSPACING_MARK},
{0x0005c6, 0x0005c6, PG_U_OTHER_PUNCTUATION},
{0x0005c7, 0x0005c7, PG_U_NONSPACING_MARK},
- {0x0005c8, 0x0005cf, PG_U_UNASSIGNED},
{0x0005d0, 0x0005ea, PG_U_OTHER_LETTER},
- {0x0005eb, 0x0005ee, PG_U_UNASSIGNED},
{0x0005ef, 0x0005f2, PG_U_OTHER_LETTER},
{0x0005f3, 0x0005f4, PG_U_OTHER_PUNCTUATION},
- {0x0005f5, 0x0005ff, PG_U_UNASSIGNED},
{0x000600, 0x000605, PG_U_FORMAT},
{0x000606, 0x000608, PG_U_MATH_SYMBOL},
{0x000609, 0x00060a, PG_U_OTHER_PUNCTUATION},
@@ -716,17 +704,14 @@ static const pg_category_range unicode_categories[4009] =
{0x0006fd, 0x0006fe, PG_U_OTHER_SYMBOL},
{0x0006ff, 0x0006ff, PG_U_OTHER_LETTER},
{0x000700, 0x00070d, PG_U_OTHER_PUNCTUATION},
- {0x00070e, 0x00070e, PG_U_UNASSIGNED},
{0x00070f, 0x00070f, PG_U_FORMAT},
{0x000710, 0x000710, PG_U_OTHER_LETTER},
{0x000711, 0x000711, PG_U_NONSPACING_MARK},
{0x000712, 0x00072f, PG_U_OTHER_LETTER},
{0x000730, 0x00074a, PG_U_NONSPACING_MARK},
- {0x00074b, 0x00074c, PG_U_UNASSIGNED},
{0x00074d, 0x0007a5, PG_U_OTHER_LETTER},
{0x0007a6, 0x0007b0, PG_U_NONSPACING_MARK},
{0x0007b1, 0x0007b1, PG_U_OTHER_LETTER},
- {0x0007b2, 0x0007bf, PG_U_UNASSIGNED},
{0x0007c0, 0x0007c9, PG_U_DECIMAL_NUMBER},
{0x0007ca, 0x0007ea, PG_U_OTHER_LETTER},
{0x0007eb, 0x0007f3, PG_U_NONSPACING_MARK},
@@ -734,7 +719,6 @@ static const pg_category_range unicode_categories[4009] =
{0x0007f6, 0x0007f6, PG_U_OTHER_SYMBOL},
{0x0007f7, 0x0007f9, PG_U_OTHER_PUNCTUATION},
{0x0007fa, 0x0007fa, PG_U_MODIFIER_LETTER},
- {0x0007fb, 0x0007fc, PG_U_UNASSIGNED},
{0x0007fd, 0x0007fd, PG_U_NONSPACING_MARK},
{0x0007fe, 0x0007ff, PG_U_CURRENCY_SYMBOL},
{0x000800, 0x000815, PG_U_OTHER_LETTER},
@@ -745,22 +729,15 @@ static const pg_category_range unicode_categories[4009] =
{0x000825, 0x000827, PG_U_NONSPACING_MARK},
{0x000828, 0x000828, PG_U_MODIFIER_LETTER},
{0x000829, 0x00082d, PG_U_NONSPACING_MARK},
- {0x00082e, 0x00082f, PG_U_UNASSIGNED},
{0x000830, 0x00083e, PG_U_OTHER_PUNCTUATION},
- {0x00083f, 0x00083f, PG_U_UNASSIGNED},
{0x000840, 0x000858, PG_U_OTHER_LETTER},
{0x000859, 0x00085b, PG_U_NONSPACING_MARK},
- {0x00085c, 0x00085d, PG_U_UNASSIGNED},
{0x00085e, 0x00085e, PG_U_OTHER_PUNCTUATION},
- {0x00085f, 0x00085f, PG_U_UNASSIGNED},
{0x000860, 0x00086a, PG_U_OTHER_LETTER},
- {0x00086b, 0x00086f, PG_U_UNASSIGNED},
{0x000870, 0x000887, PG_U_OTHER_LETTER},
{0x000888, 0x000888, PG_U_MODIFIER_SYMBOL},
{0x000889, 0x00088e, PG_U_OTHER_LETTER},
- {0x00088f, 0x00088f, PG_U_UNASSIGNED},
{0x000890, 0x000891, PG_U_FORMAT},
- {0x000892, 0x000897, PG_U_UNASSIGNED},
{0x000898, 0x00089f, PG_U_NONSPACING_MARK},
{0x0008a0, 0x0008c8, PG_U_OTHER_LETTER},
{0x0008c9, 0x0008c9, PG_U_MODIFIER_LETTER},
@@ -789,37 +766,24 @@ static const pg_category_range unicode_categories[4009] =
{0x000972, 0x000980, PG_U_OTHER_LETTER},
{0x000981, 0x000981, PG_U_NONSPACING_MARK},
{0x000982, 0x000983, PG_U_SPACING_MARK},
- {0x000984, 0x000984, PG_U_UNASSIGNED},
{0x000985, 0x00098c, PG_U_OTHER_LETTER},
- {0x00098d, 0x00098e, PG_U_UNASSIGNED},
{0x00098f, 0x000990, PG_U_OTHER_LETTER},
- {0x000991, 0x000992, PG_U_UNASSIGNED},
{0x000993, 0x0009a8, PG_U_OTHER_LETTER},
- {0x0009a9, 0x0009a9, PG_U_UNASSIGNED},
{0x0009aa, 0x0009b0, PG_U_OTHER_LETTER},
- {0x0009b1, 0x0009b1, PG_U_UNASSIGNED},
{0x0009b2, 0x0009b2, PG_U_OTHER_LETTER},
- {0x0009b3, 0x0009b5, PG_U_UNASSIGNED},
{0x0009b6, 0x0009b9, PG_U_OTHER_LETTER},
- {0x0009ba, 0x0009bb, PG_U_UNASSIGNED},
{0x0009bc, 0x0009bc, PG_U_NONSPACING_MARK},
{0x0009bd, 0x0009bd, PG_U_OTHER_LETTER},
{0x0009be, 0x0009c0, PG_U_SPACING_MARK},
{0x0009c1, 0x0009c4, PG_U_NONSPACING_MARK},
- {0x0009c5, 0x0009c6, PG_U_UNASSIGNED},
{0x0009c7, 0x0009c8, PG_U_SPACING_MARK},
- {0x0009c9, 0x0009ca, PG_U_UNASSIGNED},
{0x0009cb, 0x0009cc, PG_U_SPACING_MARK},
{0x0009cd, 0x0009cd, PG_U_NONSPACING_MARK},
{0x0009ce, 0x0009ce, PG_U_OTHER_LETTER},
- {0x0009cf, 0x0009d6, PG_U_UNASSIGNED},
{0x0009d7, 0x0009d7, PG_U_SPACING_MARK},
- {0x0009d8, 0x0009db, PG_U_UNASSIGNED},
{0x0009dc, 0x0009dd, PG_U_OTHER_LETTER},
- {0x0009de, 0x0009de, PG_U_UNASSIGNED},
{0x0009df, 0x0009e1, PG_U_OTHER_LETTER},
{0x0009e2, 0x0009e3, PG_U_NONSPACING_MARK},
- {0x0009e4, 0x0009e5, PG_U_UNASSIGNED},
{0x0009e6, 0x0009ef, PG_U_DECIMAL_NUMBER},
{0x0009f0, 0x0009f1, PG_U_OTHER_LETTER},
{0x0009f2, 0x0009f3, PG_U_CURRENCY_SYMBOL},
@@ -829,194 +793,121 @@ static const pg_category_range unicode_categories[4009] =
{0x0009fc, 0x0009fc, PG_U_OTHER_LETTER},
{0x0009fd, 0x0009fd, PG_U_OTHER_PUNCTUATION},
{0x0009fe, 0x0009fe, PG_U_NONSPACING_MARK},
- {0x0009ff, 0x000a00, PG_U_UNASSIGNED},
{0x000a01, 0x000a02, PG_U_NONSPACING_MARK},
{0x000a03, 0x000a03, PG_U_SPACING_MARK},
- {0x000a04, 0x000a04, PG_U_UNASSIGNED},
{0x000a05, 0x000a0a, PG_U_OTHER_LETTER},
- {0x000a0b, 0x000a0e, PG_U_UNASSIGNED},
{0x000a0f, 0x000a10, PG_U_OTHER_LETTER},
- {0x000a11, 0x000a12, PG_U_UNASSIGNED},
{0x000a13, 0x000a28, PG_U_OTHER_LETTER},
- {0x000a29, 0x000a29, PG_U_UNASSIGNED},
{0x000a2a, 0x000a30, PG_U_OTHER_LETTER},
- {0x000a31, 0x000a31, PG_U_UNASSIGNED},
{0x000a32, 0x000a33, PG_U_OTHER_LETTER},
- {0x000a34, 0x000a34, PG_U_UNASSIGNED},
{0x000a35, 0x000a36, PG_U_OTHER_LETTER},
- {0x000a37, 0x000a37, PG_U_UNASSIGNED},
{0x000a38, 0x000a39, PG_U_OTHER_LETTER},
- {0x000a3a, 0x000a3b, PG_U_UNASSIGNED},
{0x000a3c, 0x000a3c, PG_U_NONSPACING_MARK},
- {0x000a3d, 0x000a3d, PG_U_UNASSIGNED},
{0x000a3e, 0x000a40, PG_U_SPACING_MARK},
{0x000a41, 0x000a42, PG_U_NONSPACING_MARK},
- {0x000a43, 0x000a46, PG_U_UNASSIGNED},
{0x000a47, 0x000a48, PG_U_NONSPACING_MARK},
- {0x000a49, 0x000a4a, PG_U_UNASSIGNED},
{0x000a4b, 0x000a4d, PG_U_NONSPACING_MARK},
- {0x000a4e, 0x000a50, PG_U_UNASSIGNED},
{0x000a51, 0x000a51, PG_U_NONSPACING_MARK},
- {0x000a52, 0x000a58, PG_U_UNASSIGNED},
{0x000a59, 0x000a5c, PG_U_OTHER_LETTER},
- {0x000a5d, 0x000a5d, PG_U_UNASSIGNED},
{0x000a5e, 0x000a5e, PG_U_OTHER_LETTER},
- {0x000a5f, 0x000a65, PG_U_UNASSIGNED},
{0x000a66, 0x000a6f, PG_U_DECIMAL_NUMBER},
{0x000a70, 0x000a71, PG_U_NONSPACING_MARK},
{0x000a72, 0x000a74, PG_U_OTHER_LETTER},
{0x000a75, 0x000a75, PG_U_NONSPACING_MARK},
{0x000a76, 0x000a76, PG_U_OTHER_PUNCTUATION},
- {0x000a77, 0x000a80, PG_U_UNASSIGNED},
{0x000a81, 0x000a82, PG_U_NONSPACING_MARK},
{0x000a83, 0x000a83, PG_U_SPACING_MARK},
- {0x000a84, 0x000a84, PG_U_UNASSIGNED},
{0x000a85, 0x000a8d, PG_U_OTHER_LETTER},
- {0x000a8e, 0x000a8e, PG_U_UNASSIGNED},
{0x000a8f, 0x000a91, PG_U_OTHER_LETTER},
- {0x000a92, 0x000a92, PG_U_UNASSIGNED},
{0x000a93, 0x000aa8, PG_U_OTHER_LETTER},
- {0x000aa9, 0x000aa9, PG_U_UNASSIGNED},
{0x000aaa, 0x000ab0, PG_U_OTHER_LETTER},
- {0x000ab1, 0x000ab1, PG_U_UNASSIGNED},
{0x000ab2, 0x000ab3, PG_U_OTHER_LETTER},
- {0x000ab4, 0x000ab4, PG_U_UNASSIGNED},
{0x000ab5, 0x000ab9, PG_U_OTHER_LETTER},
- {0x000aba, 0x000abb, PG_U_UNASSIGNED},
{0x000abc, 0x000abc, PG_U_NONSPACING_MARK},
{0x000abd, 0x000abd, PG_U_OTHER_LETTER},
{0x000abe, 0x000ac0, PG_U_SPACING_MARK},
{0x000ac1, 0x000ac5, PG_U_NONSPACING_MARK},
- {0x000ac6, 0x000ac6, PG_U_UNASSIGNED},
{0x000ac7, 0x000ac8, PG_U_NONSPACING_MARK},
{0x000ac9, 0x000ac9, PG_U_SPACING_MARK},
- {0x000aca, 0x000aca, PG_U_UNASSIGNED},
{0x000acb, 0x000acc, PG_U_SPACING_MARK},
{0x000acd, 0x000acd, PG_U_NONSPACING_MARK},
- {0x000ace, 0x000acf, PG_U_UNASSIGNED},
{0x000ad0, 0x000ad0, PG_U_OTHER_LETTER},
- {0x000ad1, 0x000adf, PG_U_UNASSIGNED},
{0x000ae0, 0x000ae1, PG_U_OTHER_LETTER},
{0x000ae2, 0x000ae3, PG_U_NONSPACING_MARK},
- {0x000ae4, 0x000ae5, PG_U_UNASSIGNED},
{0x000ae6, 0x000aef, PG_U_DECIMAL_NUMBER},
{0x000af0, 0x000af0, PG_U_OTHER_PUNCTUATION},
{0x000af1, 0x000af1, PG_U_CURRENCY_SYMBOL},
- {0x000af2, 0x000af8, PG_U_UNASSIGNED},
{0x000af9, 0x000af9, PG_U_OTHER_LETTER},
{0x000afa, 0x000aff, PG_U_NONSPACING_MARK},
- {0x000b00, 0x000b00, PG_U_UNASSIGNED},
{0x000b01, 0x000b01, PG_U_NONSPACING_MARK},
{0x000b02, 0x000b03, PG_U_SPACING_MARK},
- {0x000b04, 0x000b04, PG_U_UNASSIGNED},
{0x000b05, 0x000b0c, PG_U_OTHER_LETTER},
- {0x000b0d, 0x000b0e, PG_U_UNASSIGNED},
{0x000b0f, 0x000b10, PG_U_OTHER_LETTER},
- {0x000b11, 0x000b12, PG_U_UNASSIGNED},
{0x000b13, 0x000b28, PG_U_OTHER_LETTER},
- {0x000b29, 0x000b29, PG_U_UNASSIGNED},
{0x000b2a, 0x000b30, PG_U_OTHER_LETTER},
- {0x000b31, 0x000b31, PG_U_UNASSIGNED},
{0x000b32, 0x000b33, PG_U_OTHER_LETTER},
- {0x000b34, 0x000b34, PG_U_UNASSIGNED},
{0x000b35, 0x000b39, PG_U_OTHER_LETTER},
- {0x000b3a, 0x000b3b, PG_U_UNASSIGNED},
{0x000b3c, 0x000b3c, PG_U_NONSPACING_MARK},
{0x000b3d, 0x000b3d, PG_U_OTHER_LETTER},
{0x000b3e, 0x000b3e, PG_U_SPACING_MARK},
{0x000b3f, 0x000b3f, PG_U_NONSPACING_MARK},
{0x000b40, 0x000b40, PG_U_SPACING_MARK},
{0x000b41, 0x000b44, PG_U_NONSPACING_MARK},
- {0x000b45, 0x000b46, PG_U_UNASSIGNED},
{0x000b47, 0x000b48, PG_U_SPACING_MARK},
- {0x000b49, 0x000b4a, PG_U_UNASSIGNED},
{0x000b4b, 0x000b4c, PG_U_SPACING_MARK},
{0x000b4d, 0x000b4d, PG_U_NONSPACING_MARK},
- {0x000b4e, 0x000b54, PG_U_UNASSIGNED},
{0x000b55, 0x000b56, PG_U_NONSPACING_MARK},
{0x000b57, 0x000b57, PG_U_SPACING_MARK},
- {0x000b58, 0x000b5b, PG_U_UNASSIGNED},
{0x000b5c, 0x000b5d, PG_U_OTHER_LETTER},
- {0x000b5e, 0x000b5e, PG_U_UNASSIGNED},
{0x000b5f, 0x000b61, PG_U_OTHER_LETTER},
{0x000b62, 0x000b63, PG_U_NONSPACING_MARK},
- {0x000b64, 0x000b65, PG_U_UNASSIGNED},
{0x000b66, 0x000b6f, PG_U_DECIMAL_NUMBER},
{0x000b70, 0x000b70, PG_U_OTHER_SYMBOL},
{0x000b71, 0x000b71, PG_U_OTHER_LETTER},
{0x000b72, 0x000b77, PG_U_OTHER_NUMBER},
- {0x000b78, 0x000b81, PG_U_UNASSIGNED},
{0x000b82, 0x000b82, PG_U_NONSPACING_MARK},
{0x000b83, 0x000b83, PG_U_OTHER_LETTER},
- {0x000b84, 0x000b84, PG_U_UNASSIGNED},
{0x000b85, 0x000b8a, PG_U_OTHER_LETTER},
- {0x000b8b, 0x000b8d, PG_U_UNASSIGNED},
{0x000b8e, 0x000b90, PG_U_OTHER_LETTER},
- {0x000b91, 0x000b91, PG_U_UNASSIGNED},
{0x000b92, 0x000b95, PG_U_OTHER_LETTER},
- {0x000b96, 0x000b98, PG_U_UNASSIGNED},
{0x000b99, 0x000b9a, PG_U_OTHER_LETTER},
- {0x000b9b, 0x000b9b, PG_U_UNASSIGNED},
{0x000b9c, 0x000b9c, PG_U_OTHER_LETTER},
- {0x000b9d, 0x000b9d, PG_U_UNASSIGNED},
{0x000b9e, 0x000b9f, PG_U_OTHER_LETTER},
- {0x000ba0, 0x000ba2, PG_U_UNASSIGNED},
{0x000ba3, 0x000ba4, PG_U_OTHER_LETTER},
- {0x000ba5, 0x000ba7, PG_U_UNASSIGNED},
{0x000ba8, 0x000baa, PG_U_OTHER_LETTER},
- {0x000bab, 0x000bad, PG_U_UNASSIGNED},
{0x000bae, 0x000bb9, PG_U_OTHER_LETTER},
- {0x000bba, 0x000bbd, PG_U_UNASSIGNED},
{0x000bbe, 0x000bbf, PG_U_SPACING_MARK},
{0x000bc0, 0x000bc0, PG_U_NONSPACING_MARK},
{0x000bc1, 0x000bc2, PG_U_SPACING_MARK},
- {0x000bc3, 0x000bc5, PG_U_UNASSIGNED},
{0x000bc6, 0x000bc8, PG_U_SPACING_MARK},
- {0x000bc9, 0x000bc9, PG_U_UNASSIGNED},
{0x000bca, 0x000bcc, PG_U_SPACING_MARK},
{0x000bcd, 0x000bcd, PG_U_NONSPACING_MARK},
- {0x000bce, 0x000bcf, PG_U_UNASSIGNED},
{0x000bd0, 0x000bd0, PG_U_OTHER_LETTER},
- {0x000bd1, 0x000bd6, PG_U_UNASSIGNED},
{0x000bd7, 0x000bd7, PG_U_SPACING_MARK},
- {0x000bd8, 0x000be5, PG_U_UNASSIGNED},
{0x000be6, 0x000bef, PG_U_DECIMAL_NUMBER},
{0x000bf0, 0x000bf2, PG_U_OTHER_NUMBER},
{0x000bf3, 0x000bf8, PG_U_OTHER_SYMBOL},
{0x000bf9, 0x000bf9, PG_U_CURRENCY_SYMBOL},
{0x000bfa, 0x000bfa, PG_U_OTHER_SYMBOL},
- {0x000bfb, 0x000bff, PG_U_UNASSIGNED},
{0x000c00, 0x000c00, PG_U_NONSPACING_MARK},
{0x000c01, 0x000c03, PG_U_SPACING_MARK},
{0x000c04, 0x000c04, PG_U_NONSPACING_MARK},
{0x000c05, 0x000c0c, PG_U_OTHER_LETTER},
- {0x000c0d, 0x000c0d, PG_U_UNASSIGNED},
{0x000c0e, 0x000c10, PG_U_OTHER_LETTER},
- {0x000c11, 0x000c11, PG_U_UNASSIGNED},
{0x000c12, 0x000c28, PG_U_OTHER_LETTER},
- {0x000c29, 0x000c29, PG_U_UNASSIGNED},
{0x000c2a, 0x000c39, PG_U_OTHER_LETTER},
- {0x000c3a, 0x000c3b, PG_U_UNASSIGNED},
{0x000c3c, 0x000c3c, PG_U_NONSPACING_MARK},
{0x000c3d, 0x000c3d, PG_U_OTHER_LETTER},
{0x000c3e, 0x000c40, PG_U_NONSPACING_MARK},
{0x000c41, 0x000c44, PG_U_SPACING_MARK},
- {0x000c45, 0x000c45, PG_U_UNASSIGNED},
{0x000c46, 0x000c48, PG_U_NONSPACING_MARK},
- {0x000c49, 0x000c49, PG_U_UNASSIGNED},
{0x000c4a, 0x000c4d, PG_U_NONSPACING_MARK},
- {0x000c4e, 0x000c54, PG_U_UNASSIGNED},
{0x000c55, 0x000c56, PG_U_NONSPACING_MARK},
- {0x000c57, 0x000c57, PG_U_UNASSIGNED},
{0x000c58, 0x000c5a, PG_U_OTHER_LETTER},
- {0x000c5b, 0x000c5c, PG_U_UNASSIGNED},
{0x000c5d, 0x000c5d, PG_U_OTHER_LETTER},
- {0x000c5e, 0x000c5f, PG_U_UNASSIGNED},
{0x000c60, 0x000c61, PG_U_OTHER_LETTER},
{0x000c62, 0x000c63, PG_U_NONSPACING_MARK},
- {0x000c64, 0x000c65, PG_U_UNASSIGNED},
{0x000c66, 0x000c6f, PG_U_DECIMAL_NUMBER},
- {0x000c70, 0x000c76, PG_U_UNASSIGNED},
{0x000c77, 0x000c77, PG_U_OTHER_PUNCTUATION},
{0x000c78, 0x000c7e, PG_U_OTHER_NUMBER},
{0x000c7f, 0x000c7f, PG_U_OTHER_SYMBOL},
@@ -1025,101 +916,68 @@ static const pg_category_range unicode_categories[4009] =
{0x000c82, 0x000c83, PG_U_SPACING_MARK},
{0x000c84, 0x000c84, PG_U_OTHER_PUNCTUATION},
{0x000c85, 0x000c8c, PG_U_OTHER_LETTER},
- {0x000c8d, 0x000c8d, PG_U_UNASSIGNED},
{0x000c8e, 0x000c90, PG_U_OTHER_LETTER},
- {0x000c91, 0x000c91, PG_U_UNASSIGNED},
{0x000c92, 0x000ca8, PG_U_OTHER_LETTER},
- {0x000ca9, 0x000ca9, PG_U_UNASSIGNED},
{0x000caa, 0x000cb3, PG_U_OTHER_LETTER},
- {0x000cb4, 0x000cb4, PG_U_UNASSIGNED},
{0x000cb5, 0x000cb9, PG_U_OTHER_LETTER},
- {0x000cba, 0x000cbb, PG_U_UNASSIGNED},
{0x000cbc, 0x000cbc, PG_U_NONSPACING_MARK},
{0x000cbd, 0x000cbd, PG_U_OTHER_LETTER},
{0x000cbe, 0x000cbe, PG_U_SPACING_MARK},
{0x000cbf, 0x000cbf, PG_U_NONSPACING_MARK},
{0x000cc0, 0x000cc4, PG_U_SPACING_MARK},
- {0x000cc5, 0x000cc5, PG_U_UNASSIGNED},
{0x000cc6, 0x000cc6, PG_U_NONSPACING_MARK},
{0x000cc7, 0x000cc8, PG_U_SPACING_MARK},
- {0x000cc9, 0x000cc9, PG_U_UNASSIGNED},
{0x000cca, 0x000ccb, PG_U_SPACING_MARK},
{0x000ccc, 0x000ccd, PG_U_NONSPACING_MARK},
- {0x000cce, 0x000cd4, PG_U_UNASSIGNED},
{0x000cd5, 0x000cd6, PG_U_SPACING_MARK},
- {0x000cd7, 0x000cdc, PG_U_UNASSIGNED},
{0x000cdd, 0x000cde, PG_U_OTHER_LETTER},
- {0x000cdf, 0x000cdf, PG_U_UNASSIGNED},
{0x000ce0, 0x000ce1, PG_U_OTHER_LETTER},
{0x000ce2, 0x000ce3, PG_U_NONSPACING_MARK},
- {0x000ce4, 0x000ce5, PG_U_UNASSIGNED},
{0x000ce6, 0x000cef, PG_U_DECIMAL_NUMBER},
- {0x000cf0, 0x000cf0, PG_U_UNASSIGNED},
{0x000cf1, 0x000cf2, PG_U_OTHER_LETTER},
{0x000cf3, 0x000cf3, PG_U_SPACING_MARK},
- {0x000cf4, 0x000cff, PG_U_UNASSIGNED},
{0x000d00, 0x000d01, PG_U_NONSPACING_MARK},
{0x000d02, 0x000d03, PG_U_SPACING_MARK},
{0x000d04, 0x000d0c, PG_U_OTHER_LETTER},
- {0x000d0d, 0x000d0d, PG_U_UNASSIGNED},
{0x000d0e, 0x000d10, PG_U_OTHER_LETTER},
- {0x000d11, 0x000d11, PG_U_UNASSIGNED},
{0x000d12, 0x000d3a, PG_U_OTHER_LETTER},
{0x000d3b, 0x000d3c, PG_U_NONSPACING_MARK},
{0x000d3d, 0x000d3d, PG_U_OTHER_LETTER},
{0x000d3e, 0x000d40, PG_U_SPACING_MARK},
{0x000d41, 0x000d44, PG_U_NONSPACING_MARK},
- {0x000d45, 0x000d45, PG_U_UNASSIGNED},
{0x000d46, 0x000d48, PG_U_SPACING_MARK},
- {0x000d49, 0x000d49, PG_U_UNASSIGNED},
{0x000d4a, 0x000d4c, PG_U_SPACING_MARK},
{0x000d4d, 0x000d4d, PG_U_NONSPACING_MARK},
{0x000d4e, 0x000d4e, PG_U_OTHER_LETTER},
{0x000d4f, 0x000d4f, PG_U_OTHER_SYMBOL},
- {0x000d50, 0x000d53, PG_U_UNASSIGNED},
{0x000d54, 0x000d56, PG_U_OTHER_LETTER},
{0x000d57, 0x000d57, PG_U_SPACING_MARK},
{0x000d58, 0x000d5e, PG_U_OTHER_NUMBER},
{0x000d5f, 0x000d61, PG_U_OTHER_LETTER},
{0x000d62, 0x000d63, PG_U_NONSPACING_MARK},
- {0x000d64, 0x000d65, PG_U_UNASSIGNED},
{0x000d66, 0x000d6f, PG_U_DECIMAL_NUMBER},
{0x000d70, 0x000d78, PG_U_OTHER_NUMBER},
{0x000d79, 0x000d79, PG_U_OTHER_SYMBOL},
{0x000d7a, 0x000d7f, PG_U_OTHER_LETTER},
- {0x000d80, 0x000d80, PG_U_UNASSIGNED},
{0x000d81, 0x000d81, PG_U_NONSPACING_MARK},
{0x000d82, 0x000d83, PG_U_SPACING_MARK},
- {0x000d84, 0x000d84, PG_U_UNASSIGNED},
{0x000d85, 0x000d96, PG_U_OTHER_LETTER},
- {0x000d97, 0x000d99, PG_U_UNASSIGNED},
{0x000d9a, 0x000db1, PG_U_OTHER_LETTER},
- {0x000db2, 0x000db2, PG_U_UNASSIGNED},
{0x000db3, 0x000dbb, PG_U_OTHER_LETTER},
- {0x000dbc, 0x000dbc, PG_U_UNASSIGNED},
{0x000dbd, 0x000dbd, PG_U_OTHER_LETTER},
- {0x000dbe, 0x000dbf, PG_U_UNASSIGNED},
{0x000dc0, 0x000dc6, PG_U_OTHER_LETTER},
- {0x000dc7, 0x000dc9, PG_U_UNASSIGNED},
{0x000dca, 0x000dca, PG_U_NONSPACING_MARK},
- {0x000dcb, 0x000dce, PG_U_UNASSIGNED},
{0x000dcf, 0x000dd1, PG_U_SPACING_MARK},
{0x000dd2, 0x000dd4, PG_U_NONSPACING_MARK},
- {0x000dd5, 0x000dd5, PG_U_UNASSIGNED},
{0x000dd6, 0x000dd6, PG_U_NONSPACING_MARK},
- {0x000dd7, 0x000dd7, PG_U_UNASSIGNED},
{0x000dd8, 0x000ddf, PG_U_SPACING_MARK},
- {0x000de0, 0x000de5, PG_U_UNASSIGNED},
{0x000de6, 0x000def, PG_U_DECIMAL_NUMBER},
- {0x000df0, 0x000df1, PG_U_UNASSIGNED},
{0x000df2, 0x000df3, PG_U_SPACING_MARK},
{0x000df4, 0x000df4, PG_U_OTHER_PUNCTUATION},
- {0x000df5, 0x000e00, PG_U_UNASSIGNED},
{0x000e01, 0x000e30, PG_U_OTHER_LETTER},
{0x000e31, 0x000e31, PG_U_NONSPACING_MARK},
{0x000e32, 0x000e33, PG_U_OTHER_LETTER},
{0x000e34, 0x000e3a, PG_U_NONSPACING_MARK},
- {0x000e3b, 0x000e3e, PG_U_UNASSIGNED},
{0x000e3f, 0x000e3f, PG_U_CURRENCY_SYMBOL},
{0x000e40, 0x000e45, PG_U_OTHER_LETTER},
{0x000e46, 0x000e46, PG_U_MODIFIER_LETTER},
@@ -1127,33 +985,21 @@ static const pg_category_range unicode_categories[4009] =
{0x000e4f, 0x000e4f, PG_U_OTHER_PUNCTUATION},
{0x000e50, 0x000e59, PG_U_DECIMAL_NUMBER},
{0x000e5a, 0x000e5b, PG_U_OTHER_PUNCTUATION},
- {0x000e5c, 0x000e80, PG_U_UNASSIGNED},
{0x000e81, 0x000e82, PG_U_OTHER_LETTER},
- {0x000e83, 0x000e83, PG_U_UNASSIGNED},
{0x000e84, 0x000e84, PG_U_OTHER_LETTER},
- {0x000e85, 0x000e85, PG_U_UNASSIGNED},
{0x000e86, 0x000e8a, PG_U_OTHER_LETTER},
- {0x000e8b, 0x000e8b, PG_U_UNASSIGNED},
{0x000e8c, 0x000ea3, PG_U_OTHER_LETTER},
- {0x000ea4, 0x000ea4, PG_U_UNASSIGNED},
{0x000ea5, 0x000ea5, PG_U_OTHER_LETTER},
- {0x000ea6, 0x000ea6, PG_U_UNASSIGNED},
{0x000ea7, 0x000eb0, PG_U_OTHER_LETTER},
{0x000eb1, 0x000eb1, PG_U_NONSPACING_MARK},
{0x000eb2, 0x000eb3, PG_U_OTHER_LETTER},
{0x000eb4, 0x000ebc, PG_U_NONSPACING_MARK},
{0x000ebd, 0x000ebd, PG_U_OTHER_LETTER},
- {0x000ebe, 0x000ebf, PG_U_UNASSIGNED},
{0x000ec0, 0x000ec4, PG_U_OTHER_LETTER},
- {0x000ec5, 0x000ec5, PG_U_UNASSIGNED},
{0x000ec6, 0x000ec6, PG_U_MODIFIER_LETTER},
- {0x000ec7, 0x000ec7, PG_U_UNASSIGNED},
{0x000ec8, 0x000ece, PG_U_NONSPACING_MARK},
- {0x000ecf, 0x000ecf, PG_U_UNASSIGNED},
{0x000ed0, 0x000ed9, PG_U_DECIMAL_NUMBER},
- {0x000eda, 0x000edb, PG_U_UNASSIGNED},
{0x000edc, 0x000edf, PG_U_OTHER_LETTER},
- {0x000ee0, 0x000eff, PG_U_UNASSIGNED},
{0x000f00, 0x000f00, PG_U_OTHER_LETTER},
{0x000f01, 0x000f03, PG_U_OTHER_SYMBOL},
{0x000f04, 0x000f12, PG_U_OTHER_PUNCTUATION},
@@ -1176,9 +1022,7 @@ static const pg_category_range unicode_categories[4009] =
{0x000f3d, 0x000f3d, PG_U_CLOSE_PUNCTUATION},
{0x000f3e, 0x000f3f, PG_U_SPACING_MARK},
{0x000f40, 0x000f47, PG_U_OTHER_LETTER},
- {0x000f48, 0x000f48, PG_U_UNASSIGNED},
{0x000f49, 0x000f6c, PG_U_OTHER_LETTER},
- {0x000f6d, 0x000f70, PG_U_UNASSIGNED},
{0x000f71, 0x000f7e, PG_U_NONSPACING_MARK},
{0x000f7f, 0x000f7f, PG_U_SPACING_MARK},
{0x000f80, 0x000f84, PG_U_NONSPACING_MARK},
@@ -1186,18 +1030,14 @@ static const pg_category_range unicode_categories[4009] =
{0x000f86, 0x000f87, PG_U_NONSPACING_MARK},
{0x000f88, 0x000f8c, PG_U_OTHER_LETTER},
{0x000f8d, 0x000f97, PG_U_NONSPACING_MARK},
- {0x000f98, 0x000f98, PG_U_UNASSIGNED},
{0x000f99, 0x000fbc, PG_U_NONSPACING_MARK},
- {0x000fbd, 0x000fbd, PG_U_UNASSIGNED},
{0x000fbe, 0x000fc5, PG_U_OTHER_SYMBOL},
{0x000fc6, 0x000fc6, PG_U_NONSPACING_MARK},
{0x000fc7, 0x000fcc, PG_U_OTHER_SYMBOL},
- {0x000fcd, 0x000fcd, PG_U_UNASSIGNED},
{0x000fce, 0x000fcf, PG_U_OTHER_SYMBOL},
{0x000fd0, 0x000fd4, PG_U_OTHER_PUNCTUATION},
{0x000fd5, 0x000fd8, PG_U_OTHER_SYMBOL},
{0x000fd9, 0x000fda, PG_U_OTHER_PUNCTUATION},
- {0x000fdb, 0x000fff, PG_U_UNASSIGNED},
{0x001000, 0x00102a, PG_U_OTHER_LETTER},
{0x00102b, 0x00102c, PG_U_SPACING_MARK},
{0x00102d, 0x001030, PG_U_NONSPACING_MARK},
@@ -1234,58 +1074,35 @@ static const pg_category_range unicode_categories[4009] =
{0x00109d, 0x00109d, PG_U_NONSPACING_MARK},
{0x00109e, 0x00109f, PG_U_OTHER_SYMBOL},
{0x0010a0, 0x0010c5, PG_U_UPPERCASE_LETTER},
- {0x0010c6, 0x0010c6, PG_U_UNASSIGNED},
{0x0010c7, 0x0010c7, PG_U_UPPERCASE_LETTER},
- {0x0010c8, 0x0010cc, PG_U_UNASSIGNED},
{0x0010cd, 0x0010cd, PG_U_UPPERCASE_LETTER},
- {0x0010ce, 0x0010cf, PG_U_UNASSIGNED},
{0x0010d0, 0x0010fa, PG_U_LOWERCASE_LETTER},
{0x0010fb, 0x0010fb, PG_U_OTHER_PUNCTUATION},
{0x0010fc, 0x0010fc, PG_U_MODIFIER_LETTER},
{0x0010fd, 0x0010ff, PG_U_LOWERCASE_LETTER},
{0x001100, 0x001248, PG_U_OTHER_LETTER},
- {0x001249, 0x001249, PG_U_UNASSIGNED},
{0x00124a, 0x00124d, PG_U_OTHER_LETTER},
- {0x00124e, 0x00124f, PG_U_UNASSIGNED},
{0x001250, 0x001256, PG_U_OTHER_LETTER},
- {0x001257, 0x001257, PG_U_UNASSIGNED},
{0x001258, 0x001258, PG_U_OTHER_LETTER},
- {0x001259, 0x001259, PG_U_UNASSIGNED},
{0x00125a, 0x00125d, PG_U_OTHER_LETTER},
- {0x00125e, 0x00125f, PG_U_UNASSIGNED},
{0x001260, 0x001288, PG_U_OTHER_LETTER},
- {0x001289, 0x001289, PG_U_UNASSIGNED},
{0x00128a, 0x00128d, PG_U_OTHER_LETTER},
- {0x00128e, 0x00128f, PG_U_UNASSIGNED},
{0x001290, 0x0012b0, PG_U_OTHER_LETTER},
- {0x0012b1, 0x0012b1, PG_U_UNASSIGNED},
{0x0012b2, 0x0012b5, PG_U_OTHER_LETTER},
- {0x0012b6, 0x0012b7, PG_U_UNASSIGNED},
{0x0012b8, 0x0012be, PG_U_OTHER_LETTER},
- {0x0012bf, 0x0012bf, PG_U_UNASSIGNED},
{0x0012c0, 0x0012c0, PG_U_OTHER_LETTER},
- {0x0012c1, 0x0012c1, PG_U_UNASSIGNED},
{0x0012c2, 0x0012c5, PG_U_OTHER_LETTER},
- {0x0012c6, 0x0012c7, PG_U_UNASSIGNED},
{0x0012c8, 0x0012d6, PG_U_OTHER_LETTER},
- {0x0012d7, 0x0012d7, PG_U_UNASSIGNED},
{0x0012d8, 0x001310, PG_U_OTHER_LETTER},
- {0x001311, 0x001311, PG_U_UNASSIGNED},
{0x001312, 0x001315, PG_U_OTHER_LETTER},
- {0x001316, 0x001317, PG_U_UNASSIGNED},
{0x001318, 0x00135a, PG_U_OTHER_LETTER},
- {0x00135b, 0x00135c, PG_U_UNASSIGNED},
{0x00135d, 0x00135f, PG_U_NONSPACING_MARK},
{0x001360, 0x001368, PG_U_OTHER_PUNCTUATION},
{0x001369, 0x00137c, PG_U_OTHER_NUMBER},
- {0x00137d, 0x00137f, PG_U_UNASSIGNED},
{0x001380, 0x00138f, PG_U_OTHER_LETTER},
{0x001390, 0x001399, PG_U_OTHER_SYMBOL},
- {0x00139a, 0x00139f, PG_U_UNASSIGNED},
{0x0013a0, 0x0013f5, PG_U_UPPERCASE_LETTER},
- {0x0013f6, 0x0013f7, PG_U_UNASSIGNED},
{0x0013f8, 0x0013fd, PG_U_LOWERCASE_LETTER},
- {0x0013fe, 0x0013ff, PG_U_UNASSIGNED},
{0x001400, 0x001400, PG_U_DASH_PUNCTUATION},
{0x001401, 0x00166c, PG_U_OTHER_LETTER},
{0x00166d, 0x00166d, PG_U_OTHER_SYMBOL},
@@ -1295,30 +1112,22 @@ static const pg_category_range unicode_categories[4009] =
{0x001681, 0x00169a, PG_U_OTHER_LETTER},
{0x00169b, 0x00169b, PG_U_OPEN_PUNCTUATION},
{0x00169c, 0x00169c, PG_U_CLOSE_PUNCTUATION},
- {0x00169d, 0x00169f, PG_U_UNASSIGNED},
{0x0016a0, 0x0016ea, PG_U_OTHER_LETTER},
{0x0016eb, 0x0016ed, PG_U_OTHER_PUNCTUATION},
{0x0016ee, 0x0016f0, PG_U_LETTER_NUMBER},
{0x0016f1, 0x0016f8, PG_U_OTHER_LETTER},
- {0x0016f9, 0x0016ff, PG_U_UNASSIGNED},
{0x001700, 0x001711, PG_U_OTHER_LETTER},
{0x001712, 0x001714, PG_U_NONSPACING_MARK},
{0x001715, 0x001715, PG_U_SPACING_MARK},
- {0x001716, 0x00171e, PG_U_UNASSIGNED},
{0x00171f, 0x001731, PG_U_OTHER_LETTER},
{0x001732, 0x001733, PG_U_NONSPACING_MARK},
{0x001734, 0x001734, PG_U_SPACING_MARK},
{0x001735, 0x001736, PG_U_OTHER_PUNCTUATION},
- {0x001737, 0x00173f, PG_U_UNASSIGNED},
{0x001740, 0x001751, PG_U_OTHER_LETTER},
{0x001752, 0x001753, PG_U_NONSPACING_MARK},
- {0x001754, 0x00175f, PG_U_UNASSIGNED},
{0x001760, 0x00176c, PG_U_OTHER_LETTER},
- {0x00176d, 0x00176d, PG_U_UNASSIGNED},
{0x00176e, 0x001770, PG_U_OTHER_LETTER},
- {0x001771, 0x001771, PG_U_UNASSIGNED},
{0x001772, 0x001773, PG_U_NONSPACING_MARK},
- {0x001774, 0x00177f, PG_U_UNASSIGNED},
{0x001780, 0x0017b3, PG_U_OTHER_LETTER},
{0x0017b4, 0x0017b5, PG_U_NONSPACING_MARK},
{0x0017b6, 0x0017b6, PG_U_SPACING_MARK},
@@ -1333,11 +1142,8 @@ static const pg_category_range unicode_categories[4009] =
{0x0017db, 0x0017db, PG_U_CURRENCY_SYMBOL},
{0x0017dc, 0x0017dc, PG_U_OTHER_LETTER},
{0x0017dd, 0x0017dd, PG_U_NONSPACING_MARK},
- {0x0017de, 0x0017df, PG_U_UNASSIGNED},
{0x0017e0, 0x0017e9, PG_U_DECIMAL_NUMBER},
- {0x0017ea, 0x0017ef, PG_U_UNASSIGNED},
{0x0017f0, 0x0017f9, PG_U_OTHER_NUMBER},
- {0x0017fa, 0x0017ff, PG_U_UNASSIGNED},
{0x001800, 0x001805, PG_U_OTHER_PUNCTUATION},
{0x001806, 0x001806, PG_U_DASH_PUNCTUATION},
{0x001807, 0x00180a, PG_U_OTHER_PUNCTUATION},
@@ -1345,59 +1151,44 @@ static const pg_category_range unicode_categories[4009] =
{0x00180e, 0x00180e, PG_U_FORMAT},
{0x00180f, 0x00180f, PG_U_NONSPACING_MARK},
{0x001810, 0x001819, PG_U_DECIMAL_NUMBER},
- {0x00181a, 0x00181f, PG_U_UNASSIGNED},
{0x001820, 0x001842, PG_U_OTHER_LETTER},
{0x001843, 0x001843, PG_U_MODIFIER_LETTER},
{0x001844, 0x001878, PG_U_OTHER_LETTER},
- {0x001879, 0x00187f, PG_U_UNASSIGNED},
{0x001880, 0x001884, PG_U_OTHER_LETTER},
{0x001885, 0x001886, PG_U_NONSPACING_MARK},
{0x001887, 0x0018a8, PG_U_OTHER_LETTER},
{0x0018a9, 0x0018a9, PG_U_NONSPACING_MARK},
{0x0018aa, 0x0018aa, PG_U_OTHER_LETTER},
- {0x0018ab, 0x0018af, PG_U_UNASSIGNED},
{0x0018b0, 0x0018f5, PG_U_OTHER_LETTER},
- {0x0018f6, 0x0018ff, PG_U_UNASSIGNED},
{0x001900, 0x00191e, PG_U_OTHER_LETTER},
- {0x00191f, 0x00191f, PG_U_UNASSIGNED},
{0x001920, 0x001922, PG_U_NONSPACING_MARK},
{0x001923, 0x001926, PG_U_SPACING_MARK},
{0x001927, 0x001928, PG_U_NONSPACING_MARK},
{0x001929, 0x00192b, PG_U_SPACING_MARK},
- {0x00192c, 0x00192f, PG_U_UNASSIGNED},
{0x001930, 0x001931, PG_U_SPACING_MARK},
{0x001932, 0x001932, PG_U_NONSPACING_MARK},
{0x001933, 0x001938, PG_U_SPACING_MARK},
{0x001939, 0x00193b, PG_U_NONSPACING_MARK},
- {0x00193c, 0x00193f, PG_U_UNASSIGNED},
{0x001940, 0x001940, PG_U_OTHER_SYMBOL},
- {0x001941, 0x001943, PG_U_UNASSIGNED},
{0x001944, 0x001945, PG_U_OTHER_PUNCTUATION},
{0x001946, 0x00194f, PG_U_DECIMAL_NUMBER},
{0x001950, 0x00196d, PG_U_OTHER_LETTER},
- {0x00196e, 0x00196f, PG_U_UNASSIGNED},
{0x001970, 0x001974, PG_U_OTHER_LETTER},
- {0x001975, 0x00197f, PG_U_UNASSIGNED},
{0x001980, 0x0019ab, PG_U_OTHER_LETTER},
- {0x0019ac, 0x0019af, PG_U_UNASSIGNED},
{0x0019b0, 0x0019c9, PG_U_OTHER_LETTER},
- {0x0019ca, 0x0019cf, PG_U_UNASSIGNED},
{0x0019d0, 0x0019d9, PG_U_DECIMAL_NUMBER},
{0x0019da, 0x0019da, PG_U_OTHER_NUMBER},
- {0x0019db, 0x0019dd, PG_U_UNASSIGNED},
{0x0019de, 0x0019ff, PG_U_OTHER_SYMBOL},
{0x001a00, 0x001a16, PG_U_OTHER_LETTER},
{0x001a17, 0x001a18, PG_U_NONSPACING_MARK},
{0x001a19, 0x001a1a, PG_U_SPACING_MARK},
{0x001a1b, 0x001a1b, PG_U_NONSPACING_MARK},
- {0x001a1c, 0x001a1d, PG_U_UNASSIGNED},
{0x001a1e, 0x001a1f, PG_U_OTHER_PUNCTUATION},
{0x001a20, 0x001a54, PG_U_OTHER_LETTER},
{0x001a55, 0x001a55, PG_U_SPACING_MARK},
{0x001a56, 0x001a56, PG_U_NONSPACING_MARK},
{0x001a57, 0x001a57, PG_U_SPACING_MARK},
{0x001a58, 0x001a5e, PG_U_NONSPACING_MARK},
- {0x001a5f, 0x001a5f, PG_U_UNASSIGNED},
{0x001a60, 0x001a60, PG_U_NONSPACING_MARK},
{0x001a61, 0x001a61, PG_U_SPACING_MARK},
{0x001a62, 0x001a62, PG_U_NONSPACING_MARK},
@@ -1405,20 +1196,15 @@ static const pg_category_range unicode_categories[4009] =
{0x001a65, 0x001a6c, PG_U_NONSPACING_MARK},
{0x001a6d, 0x001a72, PG_U_SPACING_MARK},
{0x001a73, 0x001a7c, PG_U_NONSPACING_MARK},
- {0x001a7d, 0x001a7e, PG_U_UNASSIGNED},
{0x001a7f, 0x001a7f, PG_U_NONSPACING_MARK},
{0x001a80, 0x001a89, PG_U_DECIMAL_NUMBER},
- {0x001a8a, 0x001a8f, PG_U_UNASSIGNED},
{0x001a90, 0x001a99, PG_U_DECIMAL_NUMBER},
- {0x001a9a, 0x001a9f, PG_U_UNASSIGNED},
{0x001aa0, 0x001aa6, PG_U_OTHER_PUNCTUATION},
{0x001aa7, 0x001aa7, PG_U_MODIFIER_LETTER},
{0x001aa8, 0x001aad, PG_U_OTHER_PUNCTUATION},
- {0x001aae, 0x001aaf, PG_U_UNASSIGNED},
{0x001ab0, 0x001abd, PG_U_NONSPACING_MARK},
{0x001abe, 0x001abe, PG_U_ENCLOSING_MARK},
{0x001abf, 0x001ace, PG_U_NONSPACING_MARK},
- {0x001acf, 0x001aff, PG_U_UNASSIGNED},
{0x001b00, 0x001b03, PG_U_NONSPACING_MARK},
{0x001b04, 0x001b04, PG_U_SPACING_MARK},
{0x001b05, 0x001b33, PG_U_OTHER_LETTER},
@@ -1431,14 +1217,12 @@ static const pg_category_range unicode_categories[4009] =
{0x001b42, 0x001b42, PG_U_NONSPACING_MARK},
{0x001b43, 0x001b44, PG_U_SPACING_MARK},
{0x001b45, 0x001b4c, PG_U_OTHER_LETTER},
- {0x001b4d, 0x001b4f, PG_U_UNASSIGNED},
{0x001b50, 0x001b59, PG_U_DECIMAL_NUMBER},
{0x001b5a, 0x001b60, PG_U_OTHER_PUNCTUATION},
{0x001b61, 0x001b6a, PG_U_OTHER_SYMBOL},
{0x001b6b, 0x001b73, PG_U_NONSPACING_MARK},
{0x001b74, 0x001b7c, PG_U_OTHER_SYMBOL},
{0x001b7d, 0x001b7e, PG_U_OTHER_PUNCTUATION},
- {0x001b7f, 0x001b7f, PG_U_UNASSIGNED},
{0x001b80, 0x001b81, PG_U_NONSPACING_MARK},
{0x001b82, 0x001b82, PG_U_SPACING_MARK},
{0x001b83, 0x001ba0, PG_U_OTHER_LETTER},
@@ -1459,29 +1243,23 @@ static const pg_category_range unicode_categories[4009] =
{0x001bee, 0x001bee, PG_U_SPACING_MARK},
{0x001bef, 0x001bf1, PG_U_NONSPACING_MARK},
{0x001bf2, 0x001bf3, PG_U_SPACING_MARK},
- {0x001bf4, 0x001bfb, PG_U_UNASSIGNED},
{0x001bfc, 0x001bff, PG_U_OTHER_PUNCTUATION},
{0x001c00, 0x001c23, PG_U_OTHER_LETTER},
{0x001c24, 0x001c2b, PG_U_SPACING_MARK},
{0x001c2c, 0x001c33, PG_U_NONSPACING_MARK},
{0x001c34, 0x001c35, PG_U_SPACING_MARK},
{0x001c36, 0x001c37, PG_U_NONSPACING_MARK},
- {0x001c38, 0x001c3a, PG_U_UNASSIGNED},
{0x001c3b, 0x001c3f, PG_U_OTHER_PUNCTUATION},
{0x001c40, 0x001c49, PG_U_DECIMAL_NUMBER},
- {0x001c4a, 0x001c4c, PG_U_UNASSIGNED},
{0x001c4d, 0x001c4f, PG_U_OTHER_LETTER},
{0x001c50, 0x001c59, PG_U_DECIMAL_NUMBER},
{0x001c5a, 0x001c77, PG_U_OTHER_LETTER},
{0x001c78, 0x001c7d, PG_U_MODIFIER_LETTER},
{0x001c7e, 0x001c7f, PG_U_OTHER_PUNCTUATION},
{0x001c80, 0x001c88, PG_U_LOWERCASE_LETTER},
- {0x001c89, 0x001c8f, PG_U_UNASSIGNED},
{0x001c90, 0x001cba, PG_U_UPPERCASE_LETTER},
- {0x001cbb, 0x001cbc, PG_U_UNASSIGNED},
{0x001cbd, 0x001cbf, PG_U_UPPERCASE_LETTER},
{0x001cc0, 0x001cc7, PG_U_OTHER_PUNCTUATION},
- {0x001cc8, 0x001ccf, PG_U_UNASSIGNED},
{0x001cd0, 0x001cd2, PG_U_NONSPACING_MARK},
{0x001cd3, 0x001cd3, PG_U_OTHER_PUNCTUATION},
{0x001cd4, 0x001ce0, PG_U_NONSPACING_MARK},
@@ -1495,7 +1273,6 @@ static const pg_category_range unicode_categories[4009] =
{0x001cf7, 0x001cf7, PG_U_SPACING_MARK},
{0x001cf8, 0x001cf9, PG_U_NONSPACING_MARK},
{0x001cfa, 0x001cfa, PG_U_OTHER_LETTER},
- {0x001cfb, 0x001cff, PG_U_UNASSIGNED},
{0x001d00, 0x001d2b, PG_U_LOWERCASE_LETTER},
{0x001d2c, 0x001d6a, PG_U_MODIFIER_LETTER},
{0x001d6b, 0x001d77, PG_U_LOWERCASE_LETTER},
@@ -1753,30 +1530,21 @@ static const pg_category_range unicode_categories[4009] =
{0x001eff, 0x001f07, PG_U_LOWERCASE_LETTER},
{0x001f08, 0x001f0f, PG_U_UPPERCASE_LETTER},
{0x001f10, 0x001f15, PG_U_LOWERCASE_LETTER},
- {0x001f16, 0x001f17, PG_U_UNASSIGNED},
{0x001f18, 0x001f1d, PG_U_UPPERCASE_LETTER},
- {0x001f1e, 0x001f1f, PG_U_UNASSIGNED},
{0x001f20, 0x001f27, PG_U_LOWERCASE_LETTER},
{0x001f28, 0x001f2f, PG_U_UPPERCASE_LETTER},
{0x001f30, 0x001f37, PG_U_LOWERCASE_LETTER},
{0x001f38, 0x001f3f, PG_U_UPPERCASE_LETTER},
{0x001f40, 0x001f45, PG_U_LOWERCASE_LETTER},
- {0x001f46, 0x001f47, PG_U_UNASSIGNED},
{0x001f48, 0x001f4d, PG_U_UPPERCASE_LETTER},
- {0x001f4e, 0x001f4f, PG_U_UNASSIGNED},
{0x001f50, 0x001f57, PG_U_LOWERCASE_LETTER},
- {0x001f58, 0x001f58, PG_U_UNASSIGNED},
{0x001f59, 0x001f59, PG_U_UPPERCASE_LETTER},
- {0x001f5a, 0x001f5a, PG_U_UNASSIGNED},
{0x001f5b, 0x001f5b, PG_U_UPPERCASE_LETTER},
- {0x001f5c, 0x001f5c, PG_U_UNASSIGNED},
{0x001f5d, 0x001f5d, PG_U_UPPERCASE_LETTER},
- {0x001f5e, 0x001f5e, PG_U_UNASSIGNED},
{0x001f5f, 0x001f5f, PG_U_UPPERCASE_LETTER},
{0x001f60, 0x001f67, PG_U_LOWERCASE_LETTER},
{0x001f68, 0x001f6f, PG_U_UPPERCASE_LETTER},
{0x001f70, 0x001f7d, PG_U_LOWERCASE_LETTER},
- {0x001f7e, 0x001f7f, PG_U_UNASSIGNED},
{0x001f80, 0x001f87, PG_U_LOWERCASE_LETTER},
{0x001f88, 0x001f8f, PG_U_TITLECASE_LETTER},
{0x001f90, 0x001f97, PG_U_LOWERCASE_LETTER},
@@ -1784,7 +1552,6 @@ static const pg_category_range unicode_categories[4009] =
{0x001fa0, 0x001fa7, PG_U_LOWERCASE_LETTER},
{0x001fa8, 0x001faf, PG_U_TITLECASE_LETTER},
{0x001fb0, 0x001fb4, PG_U_LOWERCASE_LETTER},
- {0x001fb5, 0x001fb5, PG_U_UNASSIGNED},
{0x001fb6, 0x001fb7, PG_U_LOWERCASE_LETTER},
{0x001fb8, 0x001fbb, PG_U_UPPERCASE_LETTER},
{0x001fbc, 0x001fbc, PG_U_TITLECASE_LETTER},
@@ -1792,28 +1559,22 @@ static const pg_category_range unicode_categories[4009] =
{0x001fbe, 0x001fbe, PG_U_LOWERCASE_LETTER},
{0x001fbf, 0x001fc1, PG_U_MODIFIER_SYMBOL},
{0x001fc2, 0x001fc4, PG_U_LOWERCASE_LETTER},
- {0x001fc5, 0x001fc5, PG_U_UNASSIGNED},
{0x001fc6, 0x001fc7, PG_U_LOWERCASE_LETTER},
{0x001fc8, 0x001fcb, PG_U_UPPERCASE_LETTER},
{0x001fcc, 0x001fcc, PG_U_TITLECASE_LETTER},
{0x001fcd, 0x001fcf, PG_U_MODIFIER_SYMBOL},
{0x001fd0, 0x001fd3, PG_U_LOWERCASE_LETTER},
- {0x001fd4, 0x001fd5, PG_U_UNASSIGNED},
{0x001fd6, 0x001fd7, PG_U_LOWERCASE_LETTER},
{0x001fd8, 0x001fdb, PG_U_UPPERCASE_LETTER},
- {0x001fdc, 0x001fdc, PG_U_UNASSIGNED},
{0x001fdd, 0x001fdf, PG_U_MODIFIER_SYMBOL},
{0x001fe0, 0x001fe7, PG_U_LOWERCASE_LETTER},
{0x001fe8, 0x001fec, PG_U_UPPERCASE_LETTER},
{0x001fed, 0x001fef, PG_U_MODIFIER_SYMBOL},
- {0x001ff0, 0x001ff1, PG_U_UNASSIGNED},
{0x001ff2, 0x001ff4, PG_U_LOWERCASE_LETTER},
- {0x001ff5, 0x001ff5, PG_U_UNASSIGNED},
{0x001ff6, 0x001ff7, PG_U_LOWERCASE_LETTER},
{0x001ff8, 0x001ffb, PG_U_UPPERCASE_LETTER},
{0x001ffc, 0x001ffc, PG_U_TITLECASE_LETTER},
{0x001ffd, 0x001ffe, PG_U_MODIFIER_SYMBOL},
- {0x001fff, 0x001fff, PG_U_UNASSIGNED},
{0x002000, 0x00200a, PG_U_SPACE_SEPARATOR},
{0x00200b, 0x00200f, PG_U_FORMAT},
{0x002010, 0x002015, PG_U_DASH_PUNCTUATION},
@@ -1846,11 +1607,9 @@ static const pg_category_range unicode_categories[4009] =
{0x002055, 0x00205e, PG_U_OTHER_PUNCTUATION},
{0x00205f, 0x00205f, PG_U_SPACE_SEPARATOR},
{0x002060, 0x002064, PG_U_FORMAT},
- {0x002065, 0x002065, PG_U_UNASSIGNED},
{0x002066, 0x00206f, PG_U_FORMAT},
{0x002070, 0x002070, PG_U_OTHER_NUMBER},
{0x002071, 0x002071, PG_U_MODIFIER_LETTER},
- {0x002072, 0x002073, PG_U_UNASSIGNED},
{0x002074, 0x002079, PG_U_OTHER_NUMBER},
{0x00207a, 0x00207c, PG_U_MATH_SYMBOL},
{0x00207d, 0x00207d, PG_U_OPEN_PUNCTUATION},
@@ -1860,17 +1619,13 @@ static const pg_category_range unicode_categories[4009] =
{0x00208a, 0x00208c, PG_U_MATH_SYMBOL},
{0x00208d, 0x00208d, PG_U_OPEN_PUNCTUATION},
{0x00208e, 0x00208e, PG_U_CLOSE_PUNCTUATION},
- {0x00208f, 0x00208f, PG_U_UNASSIGNED},
{0x002090, 0x00209c, PG_U_MODIFIER_LETTER},
- {0x00209d, 0x00209f, PG_U_UNASSIGNED},
{0x0020a0, 0x0020c0, PG_U_CURRENCY_SYMBOL},
- {0x0020c1, 0x0020cf, PG_U_UNASSIGNED},
{0x0020d0, 0x0020dc, PG_U_NONSPACING_MARK},
{0x0020dd, 0x0020e0, PG_U_ENCLOSING_MARK},
{0x0020e1, 0x0020e1, PG_U_NONSPACING_MARK},
{0x0020e2, 0x0020e4, PG_U_ENCLOSING_MARK},
{0x0020e5, 0x0020f0, PG_U_NONSPACING_MARK},
- {0x0020f1, 0x0020ff, PG_U_UNASSIGNED},
{0x002100, 0x002101, PG_U_OTHER_SYMBOL},
{0x002102, 0x002102, PG_U_UPPERCASE_LETTER},
{0x002103, 0x002106, PG_U_OTHER_SYMBOL},
@@ -1918,7 +1673,6 @@ static const pg_category_range unicode_categories[4009] =
{0x002185, 0x002188, PG_U_LETTER_NUMBER},
{0x002189, 0x002189, PG_U_OTHER_NUMBER},
{0x00218a, 0x00218b, PG_U_OTHER_SYMBOL},
- {0x00218c, 0x00218f, PG_U_UNASSIGNED},
{0x002190, 0x002194, PG_U_MATH_SYMBOL},
{0x002195, 0x002199, PG_U_OTHER_SYMBOL},
{0x00219a, 0x00219b, PG_U_MATH_SYMBOL},
@@ -1955,9 +1709,7 @@ static const pg_category_range unicode_categories[4009] =
{0x0023b4, 0x0023db, PG_U_OTHER_SYMBOL},
{0x0023dc, 0x0023e1, PG_U_MATH_SYMBOL},
{0x0023e2, 0x002426, PG_U_OTHER_SYMBOL},
- {0x002427, 0x00243f, PG_U_UNASSIGNED},
{0x002440, 0x00244a, PG_U_OTHER_SYMBOL},
- {0x00244b, 0x00245f, PG_U_UNASSIGNED},
{0x002460, 0x00249b, PG_U_OTHER_NUMBER},
{0x00249c, 0x0024e9, PG_U_OTHER_SYMBOL},
{0x0024ea, 0x0024ff, PG_U_OTHER_NUMBER},
@@ -2039,9 +1791,7 @@ static const pg_category_range unicode_categories[4009] =
{0x002b45, 0x002b46, PG_U_OTHER_SYMBOL},
{0x002b47, 0x002b4c, PG_U_MATH_SYMBOL},
{0x002b4d, 0x002b73, PG_U_OTHER_SYMBOL},
- {0x002b74, 0x002b75, PG_U_UNASSIGNED},
{0x002b76, 0x002b95, PG_U_OTHER_SYMBOL},
- {0x002b96, 0x002b96, PG_U_UNASSIGNED},
{0x002b97, 0x002bff, PG_U_OTHER_SYMBOL},
{0x002c00, 0x002c2f, PG_U_UPPERCASE_LETTER},
{0x002c30, 0x002c5f, PG_U_LOWERCASE_LETTER},
@@ -2170,40 +1920,25 @@ static const pg_category_range unicode_categories[4009] =
{0x002cef, 0x002cf1, PG_U_NONSPACING_MARK},
{0x002cf2, 0x002cf2, PG_U_UPPERCASE_LETTER},
{0x002cf3, 0x002cf3, PG_U_LOWERCASE_LETTER},
- {0x002cf4, 0x002cf8, PG_U_UNASSIGNED},
{0x002cf9, 0x002cfc, PG_U_OTHER_PUNCTUATION},
{0x002cfd, 0x002cfd, PG_U_OTHER_NUMBER},
{0x002cfe, 0x002cff, PG_U_OTHER_PUNCTUATION},
{0x002d00, 0x002d25, PG_U_LOWERCASE_LETTER},
- {0x002d26, 0x002d26, PG_U_UNASSIGNED},
{0x002d27, 0x002d27, PG_U_LOWERCASE_LETTER},
- {0x002d28, 0x002d2c, PG_U_UNASSIGNED},
{0x002d2d, 0x002d2d, PG_U_LOWERCASE_LETTER},
- {0x002d2e, 0x002d2f, PG_U_UNASSIGNED},
{0x002d30, 0x002d67, PG_U_OTHER_LETTER},
- {0x002d68, 0x002d6e, PG_U_UNASSIGNED},
{0x002d6f, 0x002d6f, PG_U_MODIFIER_LETTER},
{0x002d70, 0x002d70, PG_U_OTHER_PUNCTUATION},
- {0x002d71, 0x002d7e, PG_U_UNASSIGNED},
{0x002d7f, 0x002d7f, PG_U_NONSPACING_MARK},
{0x002d80, 0x002d96, PG_U_OTHER_LETTER},
- {0x002d97, 0x002d9f, PG_U_UNASSIGNED},
{0x002da0, 0x002da6, PG_U_OTHER_LETTER},
- {0x002da7, 0x002da7, PG_U_UNASSIGNED},
{0x002da8, 0x002dae, PG_U_OTHER_LETTER},
- {0x002daf, 0x002daf, PG_U_UNASSIGNED},
{0x002db0, 0x002db6, PG_U_OTHER_LETTER},
- {0x002db7, 0x002db7, PG_U_UNASSIGNED},
{0x002db8, 0x002dbe, PG_U_OTHER_LETTER},
- {0x002dbf, 0x002dbf, PG_U_UNASSIGNED},
{0x002dc0, 0x002dc6, PG_U_OTHER_LETTER},
- {0x002dc7, 0x002dc7, PG_U_UNASSIGNED},
{0x002dc8, 0x002dce, PG_U_OTHER_LETTER},
- {0x002dcf, 0x002dcf, PG_U_UNASSIGNED},
{0x002dd0, 0x002dd6, PG_U_OTHER_LETTER},
- {0x002dd7, 0x002dd7, PG_U_UNASSIGNED},
{0x002dd8, 0x002dde, PG_U_OTHER_LETTER},
- {0x002ddf, 0x002ddf, PG_U_UNASSIGNED},
{0x002de0, 0x002dff, PG_U_NONSPACING_MARK},
{0x002e00, 0x002e01, PG_U_OTHER_PUNCTUATION},
{0x002e02, 0x002e02, PG_U_INITIAL_PUNCTUATION},
@@ -2254,13 +1989,9 @@ static const pg_category_range unicode_categories[4009] =
{0x002e5b, 0x002e5b, PG_U_OPEN_PUNCTUATION},
{0x002e5c, 0x002e5c, PG_U_CLOSE_PUNCTUATION},
{0x002e5d, 0x002e5d, PG_U_DASH_PUNCTUATION},
- {0x002e5e, 0x002e7f, PG_U_UNASSIGNED},
{0x002e80, 0x002e99, PG_U_OTHER_SYMBOL},
- {0x002e9a, 0x002e9a, PG_U_UNASSIGNED},
{0x002e9b, 0x002ef3, PG_U_OTHER_SYMBOL},
- {0x002ef4, 0x002eff, PG_U_UNASSIGNED},
{0x002f00, 0x002fd5, PG_U_OTHER_SYMBOL},
- {0x002fd6, 0x002fef, PG_U_UNASSIGNED},
{0x002ff0, 0x002fff, PG_U_OTHER_SYMBOL},
{0x003000, 0x003000, PG_U_SPACE_SEPARATOR},
{0x003001, 0x003003, PG_U_OTHER_PUNCTUATION},
@@ -2302,9 +2033,7 @@ static const pg_category_range unicode_categories[4009] =
{0x00303c, 0x00303c, PG_U_OTHER_LETTER},
{0x00303d, 0x00303d, PG_U_OTHER_PUNCTUATION},
{0x00303e, 0x00303f, PG_U_OTHER_SYMBOL},
- {0x003040, 0x003040, PG_U_UNASSIGNED},
{0x003041, 0x003096, PG_U_OTHER_LETTER},
- {0x003097, 0x003098, PG_U_UNASSIGNED},
{0x003099, 0x00309a, PG_U_NONSPACING_MARK},
{0x00309b, 0x00309c, PG_U_MODIFIER_SYMBOL},
{0x00309d, 0x00309e, PG_U_MODIFIER_LETTER},
@@ -2314,21 +2043,16 @@ static const pg_category_range unicode_categories[4009] =
{0x0030fb, 0x0030fb, PG_U_OTHER_PUNCTUATION},
{0x0030fc, 0x0030fe, PG_U_MODIFIER_LETTER},
{0x0030ff, 0x0030ff, PG_U_OTHER_LETTER},
- {0x003100, 0x003104, PG_U_UNASSIGNED},
{0x003105, 0x00312f, PG_U_OTHER_LETTER},
- {0x003130, 0x003130, PG_U_UNASSIGNED},
{0x003131, 0x00318e, PG_U_OTHER_LETTER},
- {0x00318f, 0x00318f, PG_U_UNASSIGNED},
{0x003190, 0x003191, PG_U_OTHER_SYMBOL},
{0x003192, 0x003195, PG_U_OTHER_NUMBER},
{0x003196, 0x00319f, PG_U_OTHER_SYMBOL},
{0x0031a0, 0x0031bf, PG_U_OTHER_LETTER},
{0x0031c0, 0x0031e3, PG_U_OTHER_SYMBOL},
- {0x0031e4, 0x0031ee, PG_U_UNASSIGNED},
{0x0031ef, 0x0031ef, PG_U_OTHER_SYMBOL},
{0x0031f0, 0x0031ff, PG_U_OTHER_LETTER},
{0x003200, 0x00321e, PG_U_OTHER_SYMBOL},
- {0x00321f, 0x00321f, PG_U_UNASSIGNED},
{0x003220, 0x003229, PG_U_OTHER_NUMBER},
{0x00322a, 0x003247, PG_U_OTHER_SYMBOL},
{0x003248, 0x00324f, PG_U_OTHER_NUMBER},
@@ -2344,9 +2068,7 @@ static const pg_category_range unicode_categories[4009] =
{0x004e00, 0x00a014, PG_U_OTHER_LETTER},
{0x00a015, 0x00a015, PG_U_MODIFIER_LETTER},
{0x00a016, 0x00a48c, PG_U_OTHER_LETTER},
- {0x00a48d, 0x00a48f, PG_U_UNASSIGNED},
{0x00a490, 0x00a4c6, PG_U_OTHER_SYMBOL},
- {0x00a4c7, 0x00a4cf, PG_U_UNASSIGNED},
{0x00a4d0, 0x00a4f7, PG_U_OTHER_LETTER},
{0x00a4f8, 0x00a4fd, PG_U_MODIFIER_LETTER},
{0x00a4fe, 0x00a4ff, PG_U_OTHER_PUNCTUATION},
@@ -2356,7 +2078,6 @@ static const pg_category_range unicode_categories[4009] =
{0x00a610, 0x00a61f, PG_U_OTHER_LETTER},
{0x00a620, 0x00a629, PG_U_DECIMAL_NUMBER},
{0x00a62a, 0x00a62b, PG_U_OTHER_LETTER},
- {0x00a62c, 0x00a63f, PG_U_UNASSIGNED},
{0x00a640, 0x00a640, PG_U_UPPERCASE_LETTER},
{0x00a641, 0x00a641, PG_U_LOWERCASE_LETTER},
{0x00a642, 0x00a642, PG_U_UPPERCASE_LETTER},
@@ -2444,7 +2165,6 @@ static const pg_category_range unicode_categories[4009] =
{0x00a6e6, 0x00a6ef, PG_U_LETTER_NUMBER},
{0x00a6f0, 0x00a6f1, PG_U_NONSPACING_MARK},
{0x00a6f2, 0x00a6f7, PG_U_OTHER_PUNCTUATION},
- {0x00a6f8, 0x00a6ff, PG_U_UNASSIGNED},
{0x00a700, 0x00a716, PG_U_MODIFIER_SYMBOL},
{0x00a717, 0x00a71f, PG_U_MODIFIER_LETTER},
{0x00a720, 0x00a721, PG_U_MODIFIER_SYMBOL},
@@ -2593,18 +2313,14 @@ static const pg_category_range unicode_categories[4009] =
{0x00a7c8, 0x00a7c8, PG_U_LOWERCASE_LETTER},
{0x00a7c9, 0x00a7c9, PG_U_UPPERCASE_LETTER},
{0x00a7ca, 0x00a7ca, PG_U_LOWERCASE_LETTER},
- {0x00a7cb, 0x00a7cf, PG_U_UNASSIGNED},
{0x00a7d0, 0x00a7d0, PG_U_UPPERCASE_LETTER},
{0x00a7d1, 0x00a7d1, PG_U_LOWERCASE_LETTER},
- {0x00a7d2, 0x00a7d2, PG_U_UNASSIGNED},
{0x00a7d3, 0x00a7d3, PG_U_LOWERCASE_LETTER},
- {0x00a7d4, 0x00a7d4, PG_U_UNASSIGNED},
{0x00a7d5, 0x00a7d5, PG_U_LOWERCASE_LETTER},
{0x00a7d6, 0x00a7d6, PG_U_UPPERCASE_LETTER},
{0x00a7d7, 0x00a7d7, PG_U_LOWERCASE_LETTER},
{0x00a7d8, 0x00a7d8, PG_U_UPPERCASE_LETTER},
{0x00a7d9, 0x00a7d9, PG_U_LOWERCASE_LETTER},
- {0x00a7da, 0x00a7f1, PG_U_UNASSIGNED},
{0x00a7f2, 0x00a7f4, PG_U_MODIFIER_LETTER},
{0x00a7f5, 0x00a7f5, PG_U_UPPERCASE_LETTER},
{0x00a7f6, 0x00a7f6, PG_U_LOWERCASE_LETTER},
@@ -2623,23 +2339,18 @@ static const pg_category_range unicode_categories[4009] =
{0x00a827, 0x00a827, PG_U_SPACING_MARK},
{0x00a828, 0x00a82b, PG_U_OTHER_SYMBOL},
{0x00a82c, 0x00a82c, PG_U_NONSPACING_MARK},
- {0x00a82d, 0x00a82f, PG_U_UNASSIGNED},
{0x00a830, 0x00a835, PG_U_OTHER_NUMBER},
{0x00a836, 0x00a837, PG_U_OTHER_SYMBOL},
{0x00a838, 0x00a838, PG_U_CURRENCY_SYMBOL},
{0x00a839, 0x00a839, PG_U_OTHER_SYMBOL},
- {0x00a83a, 0x00a83f, PG_U_UNASSIGNED},
{0x00a840, 0x00a873, PG_U_OTHER_LETTER},
{0x00a874, 0x00a877, PG_U_OTHER_PUNCTUATION},
- {0x00a878, 0x00a87f, PG_U_UNASSIGNED},
{0x00a880, 0x00a881, PG_U_SPACING_MARK},
{0x00a882, 0x00a8b3, PG_U_OTHER_LETTER},
{0x00a8b4, 0x00a8c3, PG_U_SPACING_MARK},
{0x00a8c4, 0x00a8c5, PG_U_NONSPACING_MARK},
- {0x00a8c6, 0x00a8cd, PG_U_UNASSIGNED},
{0x00a8ce, 0x00a8cf, PG_U_OTHER_PUNCTUATION},
{0x00a8d0, 0x00a8d9, PG_U_DECIMAL_NUMBER},
- {0x00a8da, 0x00a8df, PG_U_UNASSIGNED},
{0x00a8e0, 0x00a8f1, PG_U_NONSPACING_MARK},
{0x00a8f2, 0x00a8f7, PG_U_OTHER_LETTER},
{0x00a8f8, 0x00a8fa, PG_U_OTHER_PUNCTUATION},
@@ -2654,10 +2365,8 @@ static const pg_category_range unicode_categories[4009] =
{0x00a930, 0x00a946, PG_U_OTHER_LETTER},
{0x00a947, 0x00a951, PG_U_NONSPACING_MARK},
{0x00a952, 0x00a953, PG_U_SPACING_MARK},
- {0x00a954, 0x00a95e, PG_U_UNASSIGNED},
{0x00a95f, 0x00a95f, PG_U_OTHER_PUNCTUATION},
{0x00a960, 0x00a97c, PG_U_OTHER_LETTER},
- {0x00a97d, 0x00a97f, PG_U_UNASSIGNED},
{0x00a980, 0x00a982, PG_U_NONSPACING_MARK},
{0x00a983, 0x00a983, PG_U_SPACING_MARK},
{0x00a984, 0x00a9b2, PG_U_OTHER_LETTER},
@@ -2668,10 +2377,8 @@ static const pg_category_range unicode_categories[4009] =
{0x00a9bc, 0x00a9bd, PG_U_NONSPACING_MARK},
{0x00a9be, 0x00a9c0, PG_U_SPACING_MARK},
{0x00a9c1, 0x00a9cd, PG_U_OTHER_PUNCTUATION},
- {0x00a9ce, 0x00a9ce, PG_U_UNASSIGNED},
{0x00a9cf, 0x00a9cf, PG_U_MODIFIER_LETTER},
{0x00a9d0, 0x00a9d9, PG_U_DECIMAL_NUMBER},
- {0x00a9da, 0x00a9dd, PG_U_UNASSIGNED},
{0x00a9de, 0x00a9df, PG_U_OTHER_PUNCTUATION},
{0x00a9e0, 0x00a9e4, PG_U_OTHER_LETTER},
{0x00a9e5, 0x00a9e5, PG_U_NONSPACING_MARK},
@@ -2679,22 +2386,18 @@ static const pg_category_range unicode_categories[4009] =
{0x00a9e7, 0x00a9ef, PG_U_OTHER_LETTER},
{0x00a9f0, 0x00a9f9, PG_U_DECIMAL_NUMBER},
{0x00a9fa, 0x00a9fe, PG_U_OTHER_LETTER},
- {0x00a9ff, 0x00a9ff, PG_U_UNASSIGNED},
{0x00aa00, 0x00aa28, PG_U_OTHER_LETTER},
{0x00aa29, 0x00aa2e, PG_U_NONSPACING_MARK},
{0x00aa2f, 0x00aa30, PG_U_SPACING_MARK},
{0x00aa31, 0x00aa32, PG_U_NONSPACING_MARK},
{0x00aa33, 0x00aa34, PG_U_SPACING_MARK},
{0x00aa35, 0x00aa36, PG_U_NONSPACING_MARK},
- {0x00aa37, 0x00aa3f, PG_U_UNASSIGNED},
{0x00aa40, 0x00aa42, PG_U_OTHER_LETTER},
{0x00aa43, 0x00aa43, PG_U_NONSPACING_MARK},
{0x00aa44, 0x00aa4b, PG_U_OTHER_LETTER},
{0x00aa4c, 0x00aa4c, PG_U_NONSPACING_MARK},
{0x00aa4d, 0x00aa4d, PG_U_SPACING_MARK},
- {0x00aa4e, 0x00aa4f, PG_U_UNASSIGNED},
{0x00aa50, 0x00aa59, PG_U_DECIMAL_NUMBER},
- {0x00aa5a, 0x00aa5b, PG_U_UNASSIGNED},
{0x00aa5c, 0x00aa5f, PG_U_OTHER_PUNCTUATION},
{0x00aa60, 0x00aa6f, PG_U_OTHER_LETTER},
{0x00aa70, 0x00aa70, PG_U_MODIFIER_LETTER},
@@ -2715,7 +2418,6 @@ static const pg_category_range unicode_categories[4009] =
{0x00aac0, 0x00aac0, PG_U_OTHER_LETTER},
{0x00aac1, 0x00aac1, PG_U_NONSPACING_MARK},
{0x00aac2, 0x00aac2, PG_U_OTHER_LETTER},
- {0x00aac3, 0x00aada, PG_U_UNASSIGNED},
{0x00aadb, 0x00aadc, PG_U_OTHER_LETTER},
{0x00aadd, 0x00aadd, PG_U_MODIFIER_LETTER},
{0x00aade, 0x00aadf, PG_U_OTHER_PUNCTUATION},
@@ -2728,24 +2430,17 @@ static const pg_category_range unicode_categories[4009] =
{0x00aaf3, 0x00aaf4, PG_U_MODIFIER_LETTER},
{0x00aaf5, 0x00aaf5, PG_U_SPACING_MARK},
{0x00aaf6, 0x00aaf6, PG_U_NONSPACING_MARK},
- {0x00aaf7, 0x00ab00, PG_U_UNASSIGNED},
{0x00ab01, 0x00ab06, PG_U_OTHER_LETTER},
- {0x00ab07, 0x00ab08, PG_U_UNASSIGNED},
{0x00ab09, 0x00ab0e, PG_U_OTHER_LETTER},
- {0x00ab0f, 0x00ab10, PG_U_UNASSIGNED},
{0x00ab11, 0x00ab16, PG_U_OTHER_LETTER},
- {0x00ab17, 0x00ab1f, PG_U_UNASSIGNED},
{0x00ab20, 0x00ab26, PG_U_OTHER_LETTER},
- {0x00ab27, 0x00ab27, PG_U_UNASSIGNED},
{0x00ab28, 0x00ab2e, PG_U_OTHER_LETTER},
- {0x00ab2f, 0x00ab2f, PG_U_UNASSIGNED},
{0x00ab30, 0x00ab5a, PG_U_LOWERCASE_LETTER},
{0x00ab5b, 0x00ab5b, PG_U_MODIFIER_SYMBOL},
{0x00ab5c, 0x00ab5f, PG_U_MODIFIER_LETTER},
{0x00ab60, 0x00ab68, PG_U_LOWERCASE_LETTER},
{0x00ab69, 0x00ab69, PG_U_MODIFIER_LETTER},
{0x00ab6a, 0x00ab6b, PG_U_MODIFIER_SYMBOL},
- {0x00ab6c, 0x00ab6f, PG_U_UNASSIGNED},
{0x00ab70, 0x00abbf, PG_U_LOWERCASE_LETTER},
{0x00abc0, 0x00abe2, PG_U_OTHER_LETTER},
{0x00abe3, 0x00abe4, PG_U_SPACING_MARK},
@@ -2756,52 +2451,34 @@ static const pg_category_range unicode_categories[4009] =
{0x00abeb, 0x00abeb, PG_U_OTHER_PUNCTUATION},
{0x00abec, 0x00abec, PG_U_SPACING_MARK},
{0x00abed, 0x00abed, PG_U_NONSPACING_MARK},
- {0x00abee, 0x00abef, PG_U_UNASSIGNED},
{0x00abf0, 0x00abf9, PG_U_DECIMAL_NUMBER},
- {0x00abfa, 0x00abff, PG_U_UNASSIGNED},
{0x00ac00, 0x00d7a3, PG_U_OTHER_LETTER},
- {0x00d7a4, 0x00d7af, PG_U_UNASSIGNED},
{0x00d7b0, 0x00d7c6, PG_U_OTHER_LETTER},
- {0x00d7c7, 0x00d7ca, PG_U_UNASSIGNED},
{0x00d7cb, 0x00d7fb, PG_U_OTHER_LETTER},
- {0x00d7fc, 0x00d7ff, PG_U_UNASSIGNED},
{0x00d800, 0x00dfff, PG_U_SURROGATE},
{0x00e000, 0x00f8ff, PG_U_PRIVATE_USE},
{0x00f900, 0x00fa6d, PG_U_OTHER_LETTER},
- {0x00fa6e, 0x00fa6f, PG_U_UNASSIGNED},
{0x00fa70, 0x00fad9, PG_U_OTHER_LETTER},
- {0x00fada, 0x00faff, PG_U_UNASSIGNED},
{0x00fb00, 0x00fb06, PG_U_LOWERCASE_LETTER},
- {0x00fb07, 0x00fb12, PG_U_UNASSIGNED},
{0x00fb13, 0x00fb17, PG_U_LOWERCASE_LETTER},
- {0x00fb18, 0x00fb1c, PG_U_UNASSIGNED},
{0x00fb1d, 0x00fb1d, PG_U_OTHER_LETTER},
{0x00fb1e, 0x00fb1e, PG_U_NONSPACING_MARK},
{0x00fb1f, 0x00fb28, PG_U_OTHER_LETTER},
{0x00fb29, 0x00fb29, PG_U_MATH_SYMBOL},
{0x00fb2a, 0x00fb36, PG_U_OTHER_LETTER},
- {0x00fb37, 0x00fb37, PG_U_UNASSIGNED},
{0x00fb38, 0x00fb3c, PG_U_OTHER_LETTER},
- {0x00fb3d, 0x00fb3d, PG_U_UNASSIGNED},
{0x00fb3e, 0x00fb3e, PG_U_OTHER_LETTER},
- {0x00fb3f, 0x00fb3f, PG_U_UNASSIGNED},
{0x00fb40, 0x00fb41, PG_U_OTHER_LETTER},
- {0x00fb42, 0x00fb42, PG_U_UNASSIGNED},
{0x00fb43, 0x00fb44, PG_U_OTHER_LETTER},
- {0x00fb45, 0x00fb45, PG_U_UNASSIGNED},
{0x00fb46, 0x00fbb1, PG_U_OTHER_LETTER},
{0x00fbb2, 0x00fbc2, PG_U_MODIFIER_SYMBOL},
- {0x00fbc3, 0x00fbd2, PG_U_UNASSIGNED},
{0x00fbd3, 0x00fd3d, PG_U_OTHER_LETTER},
{0x00fd3e, 0x00fd3e, PG_U_CLOSE_PUNCTUATION},
{0x00fd3f, 0x00fd3f, PG_U_OPEN_PUNCTUATION},
{0x00fd40, 0x00fd4f, PG_U_OTHER_SYMBOL},
{0x00fd50, 0x00fd8f, PG_U_OTHER_LETTER},
- {0x00fd90, 0x00fd91, PG_U_UNASSIGNED},
{0x00fd92, 0x00fdc7, PG_U_OTHER_LETTER},
- {0x00fdc8, 0x00fdce, PG_U_UNASSIGNED},
{0x00fdcf, 0x00fdcf, PG_U_OTHER_SYMBOL},
- {0x00fdd0, 0x00fdef, PG_U_UNASSIGNED},
{0x00fdf0, 0x00fdfb, PG_U_OTHER_LETTER},
{0x00fdfc, 0x00fdfc, PG_U_CURRENCY_SYMBOL},
{0x00fdfd, 0x00fdff, PG_U_OTHER_SYMBOL},
@@ -2810,7 +2487,6 @@ static const pg_category_range unicode_categories[4009] =
{0x00fe17, 0x00fe17, PG_U_OPEN_PUNCTUATION},
{0x00fe18, 0x00fe18, PG_U_CLOSE_PUNCTUATION},
{0x00fe19, 0x00fe19, PG_U_OTHER_PUNCTUATION},
- {0x00fe1a, 0x00fe1f, PG_U_UNASSIGNED},
{0x00fe20, 0x00fe2f, PG_U_NONSPACING_MARK},
{0x00fe30, 0x00fe30, PG_U_OTHER_PUNCTUATION},
{0x00fe31, 0x00fe32, PG_U_DASH_PUNCTUATION},
@@ -2837,7 +2513,6 @@ static const pg_category_range unicode_categories[4009] =
{0x00fe49, 0x00fe4c, PG_U_OTHER_PUNCTUATION},
{0x00fe4d, 0x00fe4f, PG_U_CONNECTOR_PUNCTUATION},
{0x00fe50, 0x00fe52, PG_U_OTHER_PUNCTUATION},
- {0x00fe53, 0x00fe53, PG_U_UNASSIGNED},
{0x00fe54, 0x00fe57, PG_U_OTHER_PUNCTUATION},
{0x00fe58, 0x00fe58, PG_U_DASH_PUNCTUATION},
{0x00fe59, 0x00fe59, PG_U_OPEN_PUNCTUATION},
@@ -2850,17 +2525,12 @@ static const pg_category_range unicode_categories[4009] =
{0x00fe62, 0x00fe62, PG_U_MATH_SYMBOL},
{0x00fe63, 0x00fe63, PG_U_DASH_PUNCTUATION},
{0x00fe64, 0x00fe66, PG_U_MATH_SYMBOL},
- {0x00fe67, 0x00fe67, PG_U_UNASSIGNED},
{0x00fe68, 0x00fe68, PG_U_OTHER_PUNCTUATION},
{0x00fe69, 0x00fe69, PG_U_CURRENCY_SYMBOL},
{0x00fe6a, 0x00fe6b, PG_U_OTHER_PUNCTUATION},
- {0x00fe6c, 0x00fe6f, PG_U_UNASSIGNED},
{0x00fe70, 0x00fe74, PG_U_OTHER_LETTER},
- {0x00fe75, 0x00fe75, PG_U_UNASSIGNED},
{0x00fe76, 0x00fefc, PG_U_OTHER_LETTER},
- {0x00fefd, 0x00fefe, PG_U_UNASSIGNED},
{0x00feff, 0x00feff, PG_U_FORMAT},
- {0x00ff00, 0x00ff00, PG_U_UNASSIGNED},
{0x00ff01, 0x00ff03, PG_U_OTHER_PUNCTUATION},
{0x00ff04, 0x00ff04, PG_U_CURRENCY_SYMBOL},
{0x00ff05, 0x00ff07, PG_U_OTHER_PUNCTUATION},
@@ -2898,273 +2568,175 @@ static const pg_category_range unicode_categories[4009] =
{0x00ff71, 0x00ff9d, PG_U_OTHER_LETTER},
{0x00ff9e, 0x00ff9f, PG_U_MODIFIER_LETTER},
{0x00ffa0, 0x00ffbe, PG_U_OTHER_LETTER},
- {0x00ffbf, 0x00ffc1, PG_U_UNASSIGNED},
{0x00ffc2, 0x00ffc7, PG_U_OTHER_LETTER},
- {0x00ffc8, 0x00ffc9, PG_U_UNASSIGNED},
{0x00ffca, 0x00ffcf, PG_U_OTHER_LETTER},
- {0x00ffd0, 0x00ffd1, PG_U_UNASSIGNED},
{0x00ffd2, 0x00ffd7, PG_U_OTHER_LETTER},
- {0x00ffd8, 0x00ffd9, PG_U_UNASSIGNED},
{0x00ffda, 0x00ffdc, PG_U_OTHER_LETTER},
- {0x00ffdd, 0x00ffdf, PG_U_UNASSIGNED},
{0x00ffe0, 0x00ffe1, PG_U_CURRENCY_SYMBOL},
{0x00ffe2, 0x00ffe2, PG_U_MATH_SYMBOL},
{0x00ffe3, 0x00ffe3, PG_U_MODIFIER_SYMBOL},
{0x00ffe4, 0x00ffe4, PG_U_OTHER_SYMBOL},
{0x00ffe5, 0x00ffe6, PG_U_CURRENCY_SYMBOL},
- {0x00ffe7, 0x00ffe7, PG_U_UNASSIGNED},
{0x00ffe8, 0x00ffe8, PG_U_OTHER_SYMBOL},
{0x00ffe9, 0x00ffec, PG_U_MATH_SYMBOL},
{0x00ffed, 0x00ffee, PG_U_OTHER_SYMBOL},
- {0x00ffef, 0x00fff8, PG_U_UNASSIGNED},
{0x00fff9, 0x00fffb, PG_U_FORMAT},
{0x00fffc, 0x00fffd, PG_U_OTHER_SYMBOL},
- {0x00fffe, 0x00ffff, PG_U_UNASSIGNED},
{0x010000, 0x01000b, PG_U_OTHER_LETTER},
- {0x01000c, 0x01000c, PG_U_UNASSIGNED},
{0x01000d, 0x010026, PG_U_OTHER_LETTER},
- {0x010027, 0x010027, PG_U_UNASSIGNED},
{0x010028, 0x01003a, PG_U_OTHER_LETTER},
- {0x01003b, 0x01003b, PG_U_UNASSIGNED},
{0x01003c, 0x01003d, PG_U_OTHER_LETTER},
- {0x01003e, 0x01003e, PG_U_UNASSIGNED},
{0x01003f, 0x01004d, PG_U_OTHER_LETTER},
- {0x01004e, 0x01004f, PG_U_UNASSIGNED},
{0x010050, 0x01005d, PG_U_OTHER_LETTER},
- {0x01005e, 0x01007f, PG_U_UNASSIGNED},
{0x010080, 0x0100fa, PG_U_OTHER_LETTER},
- {0x0100fb, 0x0100ff, PG_U_UNASSIGNED},
{0x010100, 0x010102, PG_U_OTHER_PUNCTUATION},
- {0x010103, 0x010106, PG_U_UNASSIGNED},
{0x010107, 0x010133, PG_U_OTHER_NUMBER},
- {0x010134, 0x010136, PG_U_UNASSIGNED},
{0x010137, 0x01013f, PG_U_OTHER_SYMBOL},
{0x010140, 0x010174, PG_U_LETTER_NUMBER},
{0x010175, 0x010178, PG_U_OTHER_NUMBER},
{0x010179, 0x010189, PG_U_OTHER_SYMBOL},
{0x01018a, 0x01018b, PG_U_OTHER_NUMBER},
{0x01018c, 0x01018e, PG_U_OTHER_SYMBOL},
- {0x01018f, 0x01018f, PG_U_UNASSIGNED},
{0x010190, 0x01019c, PG_U_OTHER_SYMBOL},
- {0x01019d, 0x01019f, PG_U_UNASSIGNED},
{0x0101a0, 0x0101a0, PG_U_OTHER_SYMBOL},
- {0x0101a1, 0x0101cf, PG_U_UNASSIGNED},
{0x0101d0, 0x0101fc, PG_U_OTHER_SYMBOL},
{0x0101fd, 0x0101fd, PG_U_NONSPACING_MARK},
- {0x0101fe, 0x01027f, PG_U_UNASSIGNED},
{0x010280, 0x01029c, PG_U_OTHER_LETTER},
- {0x01029d, 0x01029f, PG_U_UNASSIGNED},
{0x0102a0, 0x0102d0, PG_U_OTHER_LETTER},
- {0x0102d1, 0x0102df, PG_U_UNASSIGNED},
{0x0102e0, 0x0102e0, PG_U_NONSPACING_MARK},
{0x0102e1, 0x0102fb, PG_U_OTHER_NUMBER},
- {0x0102fc, 0x0102ff, PG_U_UNASSIGNED},
{0x010300, 0x01031f, PG_U_OTHER_LETTER},
{0x010320, 0x010323, PG_U_OTHER_NUMBER},
- {0x010324, 0x01032c, PG_U_UNASSIGNED},
{0x01032d, 0x010340, PG_U_OTHER_LETTER},
{0x010341, 0x010341, PG_U_LETTER_NUMBER},
{0x010342, 0x010349, PG_U_OTHER_LETTER},
{0x01034a, 0x01034a, PG_U_LETTER_NUMBER},
- {0x01034b, 0x01034f, PG_U_UNASSIGNED},
{0x010350, 0x010375, PG_U_OTHER_LETTER},
{0x010376, 0x01037a, PG_U_NONSPACING_MARK},
- {0x01037b, 0x01037f, PG_U_UNASSIGNED},
{0x010380, 0x01039d, PG_U_OTHER_LETTER},
- {0x01039e, 0x01039e, PG_U_UNASSIGNED},
{0x01039f, 0x01039f, PG_U_OTHER_PUNCTUATION},
{0x0103a0, 0x0103c3, PG_U_OTHER_LETTER},
- {0x0103c4, 0x0103c7, PG_U_UNASSIGNED},
{0x0103c8, 0x0103cf, PG_U_OTHER_LETTER},
{0x0103d0, 0x0103d0, PG_U_OTHER_PUNCTUATION},
{0x0103d1, 0x0103d5, PG_U_LETTER_NUMBER},
- {0x0103d6, 0x0103ff, PG_U_UNASSIGNED},
{0x010400, 0x010427, PG_U_UPPERCASE_LETTER},
{0x010428, 0x01044f, PG_U_LOWERCASE_LETTER},
{0x010450, 0x01049d, PG_U_OTHER_LETTER},
- {0x01049e, 0x01049f, PG_U_UNASSIGNED},
{0x0104a0, 0x0104a9, PG_U_DECIMAL_NUMBER},
- {0x0104aa, 0x0104af, PG_U_UNASSIGNED},
{0x0104b0, 0x0104d3, PG_U_UPPERCASE_LETTER},
- {0x0104d4, 0x0104d7, PG_U_UNASSIGNED},
{0x0104d8, 0x0104fb, PG_U_LOWERCASE_LETTER},
- {0x0104fc, 0x0104ff, PG_U_UNASSIGNED},
{0x010500, 0x010527, PG_U_OTHER_LETTER},
- {0x010528, 0x01052f, PG_U_UNASSIGNED},
{0x010530, 0x010563, PG_U_OTHER_LETTER},
- {0x010564, 0x01056e, PG_U_UNASSIGNED},
{0x01056f, 0x01056f, PG_U_OTHER_PUNCTUATION},
{0x010570, 0x01057a, PG_U_UPPERCASE_LETTER},
- {0x01057b, 0x01057b, PG_U_UNASSIGNED},
{0x01057c, 0x01058a, PG_U_UPPERCASE_LETTER},
- {0x01058b, 0x01058b, PG_U_UNASSIGNED},
{0x01058c, 0x010592, PG_U_UPPERCASE_LETTER},
- {0x010593, 0x010593, PG_U_UNASSIGNED},
{0x010594, 0x010595, PG_U_UPPERCASE_LETTER},
- {0x010596, 0x010596, PG_U_UNASSIGNED},
{0x010597, 0x0105a1, PG_U_LOWERCASE_LETTER},
- {0x0105a2, 0x0105a2, PG_U_UNASSIGNED},
{0x0105a3, 0x0105b1, PG_U_LOWERCASE_LETTER},
- {0x0105b2, 0x0105b2, PG_U_UNASSIGNED},
{0x0105b3, 0x0105b9, PG_U_LOWERCASE_LETTER},
- {0x0105ba, 0x0105ba, PG_U_UNASSIGNED},
{0x0105bb, 0x0105bc, PG_U_LOWERCASE_LETTER},
- {0x0105bd, 0x0105ff, PG_U_UNASSIGNED},
{0x010600, 0x010736, PG_U_OTHER_LETTER},
- {0x010737, 0x01073f, PG_U_UNASSIGNED},
{0x010740, 0x010755, PG_U_OTHER_LETTER},
- {0x010756, 0x01075f, PG_U_UNASSIGNED},
{0x010760, 0x010767, PG_U_OTHER_LETTER},
- {0x010768, 0x01077f, PG_U_UNASSIGNED},
{0x010780, 0x010785, PG_U_MODIFIER_LETTER},
- {0x010786, 0x010786, PG_U_UNASSIGNED},
{0x010787, 0x0107b0, PG_U_MODIFIER_LETTER},
- {0x0107b1, 0x0107b1, PG_U_UNASSIGNED},
{0x0107b2, 0x0107ba, PG_U_MODIFIER_LETTER},
- {0x0107bb, 0x0107ff, PG_U_UNASSIGNED},
{0x010800, 0x010805, PG_U_OTHER_LETTER},
- {0x010806, 0x010807, PG_U_UNASSIGNED},
{0x010808, 0x010808, PG_U_OTHER_LETTER},
- {0x010809, 0x010809, PG_U_UNASSIGNED},
{0x01080a, 0x010835, PG_U_OTHER_LETTER},
- {0x010836, 0x010836, PG_U_UNASSIGNED},
{0x010837, 0x010838, PG_U_OTHER_LETTER},
- {0x010839, 0x01083b, PG_U_UNASSIGNED},
{0x01083c, 0x01083c, PG_U_OTHER_LETTER},
- {0x01083d, 0x01083e, PG_U_UNASSIGNED},
{0x01083f, 0x010855, PG_U_OTHER_LETTER},
- {0x010856, 0x010856, PG_U_UNASSIGNED},
{0x010857, 0x010857, PG_U_OTHER_PUNCTUATION},
{0x010858, 0x01085f, PG_U_OTHER_NUMBER},
{0x010860, 0x010876, PG_U_OTHER_LETTER},
{0x010877, 0x010878, PG_U_OTHER_SYMBOL},
{0x010879, 0x01087f, PG_U_OTHER_NUMBER},
{0x010880, 0x01089e, PG_U_OTHER_LETTER},
- {0x01089f, 0x0108a6, PG_U_UNASSIGNED},
{0x0108a7, 0x0108af, PG_U_OTHER_NUMBER},
- {0x0108b0, 0x0108df, PG_U_UNASSIGNED},
{0x0108e0, 0x0108f2, PG_U_OTHER_LETTER},
- {0x0108f3, 0x0108f3, PG_U_UNASSIGNED},
{0x0108f4, 0x0108f5, PG_U_OTHER_LETTER},
- {0x0108f6, 0x0108fa, PG_U_UNASSIGNED},
{0x0108fb, 0x0108ff, PG_U_OTHER_NUMBER},
{0x010900, 0x010915, PG_U_OTHER_LETTER},
{0x010916, 0x01091b, PG_U_OTHER_NUMBER},
- {0x01091c, 0x01091e, PG_U_UNASSIGNED},
{0x01091f, 0x01091f, PG_U_OTHER_PUNCTUATION},
{0x010920, 0x010939, PG_U_OTHER_LETTER},
- {0x01093a, 0x01093e, PG_U_UNASSIGNED},
{0x01093f, 0x01093f, PG_U_OTHER_PUNCTUATION},
- {0x010940, 0x01097f, PG_U_UNASSIGNED},
{0x010980, 0x0109b7, PG_U_OTHER_LETTER},
- {0x0109b8, 0x0109bb, PG_U_UNASSIGNED},
{0x0109bc, 0x0109bd, PG_U_OTHER_NUMBER},
{0x0109be, 0x0109bf, PG_U_OTHER_LETTER},
{0x0109c0, 0x0109cf, PG_U_OTHER_NUMBER},
- {0x0109d0, 0x0109d1, PG_U_UNASSIGNED},
{0x0109d2, 0x0109ff, PG_U_OTHER_NUMBER},
{0x010a00, 0x010a00, PG_U_OTHER_LETTER},
{0x010a01, 0x010a03, PG_U_NONSPACING_MARK},
- {0x010a04, 0x010a04, PG_U_UNASSIGNED},
{0x010a05, 0x010a06, PG_U_NONSPACING_MARK},
- {0x010a07, 0x010a0b, PG_U_UNASSIGNED},
{0x010a0c, 0x010a0f, PG_U_NONSPACING_MARK},
{0x010a10, 0x010a13, PG_U_OTHER_LETTER},
- {0x010a14, 0x010a14, PG_U_UNASSIGNED},
{0x010a15, 0x010a17, PG_U_OTHER_LETTER},
- {0x010a18, 0x010a18, PG_U_UNASSIGNED},
{0x010a19, 0x010a35, PG_U_OTHER_LETTER},
- {0x010a36, 0x010a37, PG_U_UNASSIGNED},
{0x010a38, 0x010a3a, PG_U_NONSPACING_MARK},
- {0x010a3b, 0x010a3e, PG_U_UNASSIGNED},
{0x010a3f, 0x010a3f, PG_U_NONSPACING_MARK},
{0x010a40, 0x010a48, PG_U_OTHER_NUMBER},
- {0x010a49, 0x010a4f, PG_U_UNASSIGNED},
{0x010a50, 0x010a58, PG_U_OTHER_PUNCTUATION},
- {0x010a59, 0x010a5f, PG_U_UNASSIGNED},
{0x010a60, 0x010a7c, PG_U_OTHER_LETTER},
{0x010a7d, 0x010a7e, PG_U_OTHER_NUMBER},
{0x010a7f, 0x010a7f, PG_U_OTHER_PUNCTUATION},
{0x010a80, 0x010a9c, PG_U_OTHER_LETTER},
{0x010a9d, 0x010a9f, PG_U_OTHER_NUMBER},
- {0x010aa0, 0x010abf, PG_U_UNASSIGNED},
{0x010ac0, 0x010ac7, PG_U_OTHER_LETTER},
{0x010ac8, 0x010ac8, PG_U_OTHER_SYMBOL},
{0x010ac9, 0x010ae4, PG_U_OTHER_LETTER},
{0x010ae5, 0x010ae6, PG_U_NONSPACING_MARK},
- {0x010ae7, 0x010aea, PG_U_UNASSIGNED},
{0x010aeb, 0x010aef, PG_U_OTHER_NUMBER},
{0x010af0, 0x010af6, PG_U_OTHER_PUNCTUATION},
- {0x010af7, 0x010aff, PG_U_UNASSIGNED},
{0x010b00, 0x010b35, PG_U_OTHER_LETTER},
- {0x010b36, 0x010b38, PG_U_UNASSIGNED},
{0x010b39, 0x010b3f, PG_U_OTHER_PUNCTUATION},
{0x010b40, 0x010b55, PG_U_OTHER_LETTER},
- {0x010b56, 0x010b57, PG_U_UNASSIGNED},
{0x010b58, 0x010b5f, PG_U_OTHER_NUMBER},
{0x010b60, 0x010b72, PG_U_OTHER_LETTER},
- {0x010b73, 0x010b77, PG_U_UNASSIGNED},
{0x010b78, 0x010b7f, PG_U_OTHER_NUMBER},
{0x010b80, 0x010b91, PG_U_OTHER_LETTER},
- {0x010b92, 0x010b98, PG_U_UNASSIGNED},
{0x010b99, 0x010b9c, PG_U_OTHER_PUNCTUATION},
- {0x010b9d, 0x010ba8, PG_U_UNASSIGNED},
{0x010ba9, 0x010baf, PG_U_OTHER_NUMBER},
- {0x010bb0, 0x010bff, PG_U_UNASSIGNED},
{0x010c00, 0x010c48, PG_U_OTHER_LETTER},
- {0x010c49, 0x010c7f, PG_U_UNASSIGNED},
{0x010c80, 0x010cb2, PG_U_UPPERCASE_LETTER},
- {0x010cb3, 0x010cbf, PG_U_UNASSIGNED},
{0x010cc0, 0x010cf2, PG_U_LOWERCASE_LETTER},
- {0x010cf3, 0x010cf9, PG_U_UNASSIGNED},
{0x010cfa, 0x010cff, PG_U_OTHER_NUMBER},
{0x010d00, 0x010d23, PG_U_OTHER_LETTER},
{0x010d24, 0x010d27, PG_U_NONSPACING_MARK},
- {0x010d28, 0x010d2f, PG_U_UNASSIGNED},
{0x010d30, 0x010d39, PG_U_DECIMAL_NUMBER},
- {0x010d3a, 0x010e5f, PG_U_UNASSIGNED},
{0x010e60, 0x010e7e, PG_U_OTHER_NUMBER},
- {0x010e7f, 0x010e7f, PG_U_UNASSIGNED},
{0x010e80, 0x010ea9, PG_U_OTHER_LETTER},
- {0x010eaa, 0x010eaa, PG_U_UNASSIGNED},
{0x010eab, 0x010eac, PG_U_NONSPACING_MARK},
{0x010ead, 0x010ead, PG_U_DASH_PUNCTUATION},
- {0x010eae, 0x010eaf, PG_U_UNASSIGNED},
{0x010eb0, 0x010eb1, PG_U_OTHER_LETTER},
- {0x010eb2, 0x010efc, PG_U_UNASSIGNED},
{0x010efd, 0x010eff, PG_U_NONSPACING_MARK},
{0x010f00, 0x010f1c, PG_U_OTHER_LETTER},
{0x010f1d, 0x010f26, PG_U_OTHER_NUMBER},
{0x010f27, 0x010f27, PG_U_OTHER_LETTER},
- {0x010f28, 0x010f2f, PG_U_UNASSIGNED},
{0x010f30, 0x010f45, PG_U_OTHER_LETTER},
{0x010f46, 0x010f50, PG_U_NONSPACING_MARK},
{0x010f51, 0x010f54, PG_U_OTHER_NUMBER},
{0x010f55, 0x010f59, PG_U_OTHER_PUNCTUATION},
- {0x010f5a, 0x010f6f, PG_U_UNASSIGNED},
{0x010f70, 0x010f81, PG_U_OTHER_LETTER},
{0x010f82, 0x010f85, PG_U_NONSPACING_MARK},
{0x010f86, 0x010f89, PG_U_OTHER_PUNCTUATION},
- {0x010f8a, 0x010faf, PG_U_UNASSIGNED},
{0x010fb0, 0x010fc4, PG_U_OTHER_LETTER},
{0x010fc5, 0x010fcb, PG_U_OTHER_NUMBER},
- {0x010fcc, 0x010fdf, PG_U_UNASSIGNED},
{0x010fe0, 0x010ff6, PG_U_OTHER_LETTER},
- {0x010ff7, 0x010fff, PG_U_UNASSIGNED},
{0x011000, 0x011000, PG_U_SPACING_MARK},
{0x011001, 0x011001, PG_U_NONSPACING_MARK},
{0x011002, 0x011002, PG_U_SPACING_MARK},
{0x011003, 0x011037, PG_U_OTHER_LETTER},
{0x011038, 0x011046, PG_U_NONSPACING_MARK},
{0x011047, 0x01104d, PG_U_OTHER_PUNCTUATION},
- {0x01104e, 0x011051, PG_U_UNASSIGNED},
{0x011052, 0x011065, PG_U_OTHER_NUMBER},
{0x011066, 0x01106f, PG_U_DECIMAL_NUMBER},
{0x011070, 0x011070, PG_U_NONSPACING_MARK},
{0x011071, 0x011072, PG_U_OTHER_LETTER},
{0x011073, 0x011074, PG_U_NONSPACING_MARK},
{0x011075, 0x011075, PG_U_OTHER_LETTER},
- {0x011076, 0x01107e, PG_U_UNASSIGNED},
{0x01107f, 0x011081, PG_U_NONSPACING_MARK},
{0x011082, 0x011082, PG_U_SPACING_MARK},
{0x011083, 0x0110af, PG_U_OTHER_LETTER},
@@ -3176,30 +2748,23 @@ static const pg_category_range unicode_categories[4009] =
{0x0110bd, 0x0110bd, PG_U_FORMAT},
{0x0110be, 0x0110c1, PG_U_OTHER_PUNCTUATION},
{0x0110c2, 0x0110c2, PG_U_NONSPACING_MARK},
- {0x0110c3, 0x0110cc, PG_U_UNASSIGNED},
{0x0110cd, 0x0110cd, PG_U_FORMAT},
- {0x0110ce, 0x0110cf, PG_U_UNASSIGNED},
{0x0110d0, 0x0110e8, PG_U_OTHER_LETTER},
- {0x0110e9, 0x0110ef, PG_U_UNASSIGNED},
{0x0110f0, 0x0110f9, PG_U_DECIMAL_NUMBER},
- {0x0110fa, 0x0110ff, PG_U_UNASSIGNED},
{0x011100, 0x011102, PG_U_NONSPACING_MARK},
{0x011103, 0x011126, PG_U_OTHER_LETTER},
{0x011127, 0x01112b, PG_U_NONSPACING_MARK},
{0x01112c, 0x01112c, PG_U_SPACING_MARK},
{0x01112d, 0x011134, PG_U_NONSPACING_MARK},
- {0x011135, 0x011135, PG_U_UNASSIGNED},
{0x011136, 0x01113f, PG_U_DECIMAL_NUMBER},
{0x011140, 0x011143, PG_U_OTHER_PUNCTUATION},
{0x011144, 0x011144, PG_U_OTHER_LETTER},
{0x011145, 0x011146, PG_U_SPACING_MARK},
{0x011147, 0x011147, PG_U_OTHER_LETTER},
- {0x011148, 0x01114f, PG_U_UNASSIGNED},
{0x011150, 0x011172, PG_U_OTHER_LETTER},
{0x011173, 0x011173, PG_U_NONSPACING_MARK},
{0x011174, 0x011175, PG_U_OTHER_PUNCTUATION},
{0x011176, 0x011176, PG_U_OTHER_LETTER},
- {0x011177, 0x01117f, PG_U_UNASSIGNED},
{0x011180, 0x011181, PG_U_NONSPACING_MARK},
{0x011182, 0x011182, PG_U_SPACING_MARK},
{0x011183, 0x0111b2, PG_U_OTHER_LETTER},
@@ -3217,11 +2782,8 @@ static const pg_category_range unicode_categories[4009] =
{0x0111db, 0x0111db, PG_U_OTHER_PUNCTUATION},
{0x0111dc, 0x0111dc, PG_U_OTHER_LETTER},
{0x0111dd, 0x0111df, PG_U_OTHER_PUNCTUATION},
- {0x0111e0, 0x0111e0, PG_U_UNASSIGNED},
{0x0111e1, 0x0111f4, PG_U_OTHER_NUMBER},
- {0x0111f5, 0x0111ff, PG_U_UNASSIGNED},
{0x011200, 0x011211, PG_U_OTHER_LETTER},
- {0x011212, 0x011212, PG_U_UNASSIGNED},
{0x011213, 0x01122b, PG_U_OTHER_LETTER},
{0x01122c, 0x01122e, PG_U_SPACING_MARK},
{0x01122f, 0x011231, PG_U_NONSPACING_MARK},
@@ -3233,61 +2795,38 @@ static const pg_category_range unicode_categories[4009] =
{0x01123e, 0x01123e, PG_U_NONSPACING_MARK},
{0x01123f, 0x011240, PG_U_OTHER_LETTER},
{0x011241, 0x011241, PG_U_NONSPACING_MARK},
- {0x011242, 0x01127f, PG_U_UNASSIGNED},
{0x011280, 0x011286, PG_U_OTHER_LETTER},
- {0x011287, 0x011287, PG_U_UNASSIGNED},
{0x011288, 0x011288, PG_U_OTHER_LETTER},
- {0x011289, 0x011289, PG_U_UNASSIGNED},
{0x01128a, 0x01128d, PG_U_OTHER_LETTER},
- {0x01128e, 0x01128e, PG_U_UNASSIGNED},
{0x01128f, 0x01129d, PG_U_OTHER_LETTER},
- {0x01129e, 0x01129e, PG_U_UNASSIGNED},
{0x01129f, 0x0112a8, PG_U_OTHER_LETTER},
{0x0112a9, 0x0112a9, PG_U_OTHER_PUNCTUATION},
- {0x0112aa, 0x0112af, PG_U_UNASSIGNED},
{0x0112b0, 0x0112de, PG_U_OTHER_LETTER},
{0x0112df, 0x0112df, PG_U_NONSPACING_MARK},
{0x0112e0, 0x0112e2, PG_U_SPACING_MARK},
{0x0112e3, 0x0112ea, PG_U_NONSPACING_MARK},
- {0x0112eb, 0x0112ef, PG_U_UNASSIGNED},
{0x0112f0, 0x0112f9, PG_U_DECIMAL_NUMBER},
- {0x0112fa, 0x0112ff, PG_U_UNASSIGNED},
{0x011300, 0x011301, PG_U_NONSPACING_MARK},
{0x011302, 0x011303, PG_U_SPACING_MARK},
- {0x011304, 0x011304, PG_U_UNASSIGNED},
{0x011305, 0x01130c, PG_U_OTHER_LETTER},
- {0x01130d, 0x01130e, PG_U_UNASSIGNED},
{0x01130f, 0x011310, PG_U_OTHER_LETTER},
- {0x011311, 0x011312, PG_U_UNASSIGNED},
{0x011313, 0x011328, PG_U_OTHER_LETTER},
- {0x011329, 0x011329, PG_U_UNASSIGNED},
{0x01132a, 0x011330, PG_U_OTHER_LETTER},
- {0x011331, 0x011331, PG_U_UNASSIGNED},
{0x011332, 0x011333, PG_U_OTHER_LETTER},
- {0x011334, 0x011334, PG_U_UNASSIGNED},
{0x011335, 0x011339, PG_U_OTHER_LETTER},
- {0x01133a, 0x01133a, PG_U_UNASSIGNED},
{0x01133b, 0x01133c, PG_U_NONSPACING_MARK},
{0x01133d, 0x01133d, PG_U_OTHER_LETTER},
{0x01133e, 0x01133f, PG_U_SPACING_MARK},
{0x011340, 0x011340, PG_U_NONSPACING_MARK},
{0x011341, 0x011344, PG_U_SPACING_MARK},
- {0x011345, 0x011346, PG_U_UNASSIGNED},
{0x011347, 0x011348, PG_U_SPACING_MARK},
- {0x011349, 0x01134a, PG_U_UNASSIGNED},
{0x01134b, 0x01134d, PG_U_SPACING_MARK},
- {0x01134e, 0x01134f, PG_U_UNASSIGNED},
{0x011350, 0x011350, PG_U_OTHER_LETTER},
- {0x011351, 0x011356, PG_U_UNASSIGNED},
{0x011357, 0x011357, PG_U_SPACING_MARK},
- {0x011358, 0x01135c, PG_U_UNASSIGNED},
{0x01135d, 0x011361, PG_U_OTHER_LETTER},
{0x011362, 0x011363, PG_U_SPACING_MARK},
- {0x011364, 0x011365, PG_U_UNASSIGNED},
{0x011366, 0x01136c, PG_U_NONSPACING_MARK},
- {0x01136d, 0x01136f, PG_U_UNASSIGNED},
{0x011370, 0x011374, PG_U_NONSPACING_MARK},
- {0x011375, 0x0113ff, PG_U_UNASSIGNED},
{0x011400, 0x011434, PG_U_OTHER_LETTER},
{0x011435, 0x011437, PG_U_SPACING_MARK},
{0x011438, 0x01143f, PG_U_NONSPACING_MARK},
@@ -3299,11 +2838,9 @@ static const pg_category_range unicode_categories[4009] =
{0x01144b, 0x01144f, PG_U_OTHER_PUNCTUATION},
{0x011450, 0x011459, PG_U_DECIMAL_NUMBER},
{0x01145a, 0x01145b, PG_U_OTHER_PUNCTUATION},
- {0x01145c, 0x01145c, PG_U_UNASSIGNED},
{0x01145d, 0x01145d, PG_U_OTHER_PUNCTUATION},
{0x01145e, 0x01145e, PG_U_NONSPACING_MARK},
{0x01145f, 0x011461, PG_U_OTHER_LETTER},
- {0x011462, 0x01147f, PG_U_UNASSIGNED},
{0x011480, 0x0114af, PG_U_OTHER_LETTER},
{0x0114b0, 0x0114b2, PG_U_SPACING_MARK},
{0x0114b3, 0x0114b8, PG_U_NONSPACING_MARK},
@@ -3316,13 +2853,10 @@ static const pg_category_range unicode_categories[4009] =
{0x0114c4, 0x0114c5, PG_U_OTHER_LETTER},
{0x0114c6, 0x0114c6, PG_U_OTHER_PUNCTUATION},
{0x0114c7, 0x0114c7, PG_U_OTHER_LETTER},
- {0x0114c8, 0x0114cf, PG_U_UNASSIGNED},
{0x0114d0, 0x0114d9, PG_U_DECIMAL_NUMBER},
- {0x0114da, 0x01157f, PG_U_UNASSIGNED},
{0x011580, 0x0115ae, PG_U_OTHER_LETTER},
{0x0115af, 0x0115b1, PG_U_SPACING_MARK},
{0x0115b2, 0x0115b5, PG_U_NONSPACING_MARK},
- {0x0115b6, 0x0115b7, PG_U_UNASSIGNED},
{0x0115b8, 0x0115bb, PG_U_SPACING_MARK},
{0x0115bc, 0x0115bd, PG_U_NONSPACING_MARK},
{0x0115be, 0x0115be, PG_U_SPACING_MARK},
@@ -3330,7 +2864,6 @@ static const pg_category_range unicode_categories[4009] =
{0x0115c1, 0x0115d7, PG_U_OTHER_PUNCTUATION},
{0x0115d8, 0x0115db, PG_U_OTHER_LETTER},
{0x0115dc, 0x0115dd, PG_U_NONSPACING_MARK},
- {0x0115de, 0x0115ff, PG_U_UNASSIGNED},
{0x011600, 0x01162f, PG_U_OTHER_LETTER},
{0x011630, 0x011632, PG_U_SPACING_MARK},
{0x011633, 0x01163a, PG_U_NONSPACING_MARK},
@@ -3340,11 +2873,8 @@ static const pg_category_range unicode_categories[4009] =
{0x01163f, 0x011640, PG_U_NONSPACING_MARK},
{0x011641, 0x011643, PG_U_OTHER_PUNCTUATION},
{0x011644, 0x011644, PG_U_OTHER_LETTER},
- {0x011645, 0x01164f, PG_U_UNASSIGNED},
{0x011650, 0x011659, PG_U_DECIMAL_NUMBER},
- {0x01165a, 0x01165f, PG_U_UNASSIGNED},
{0x011660, 0x01166c, PG_U_OTHER_PUNCTUATION},
- {0x01166d, 0x01167f, PG_U_UNASSIGNED},
{0x011680, 0x0116aa, PG_U_OTHER_LETTER},
{0x0116ab, 0x0116ab, PG_U_NONSPACING_MARK},
{0x0116ac, 0x0116ac, PG_U_SPACING_MARK},
@@ -3355,48 +2885,35 @@ static const pg_category_range unicode_categories[4009] =
{0x0116b7, 0x0116b7, PG_U_NONSPACING_MARK},
{0x0116b8, 0x0116b8, PG_U_OTHER_LETTER},
{0x0116b9, 0x0116b9, PG_U_OTHER_PUNCTUATION},
- {0x0116ba, 0x0116bf, PG_U_UNASSIGNED},
{0x0116c0, 0x0116c9, PG_U_DECIMAL_NUMBER},
- {0x0116ca, 0x0116ff, PG_U_UNASSIGNED},
{0x011700, 0x01171a, PG_U_OTHER_LETTER},
- {0x01171b, 0x01171c, PG_U_UNASSIGNED},
{0x01171d, 0x01171f, PG_U_NONSPACING_MARK},
{0x011720, 0x011721, PG_U_SPACING_MARK},
{0x011722, 0x011725, PG_U_NONSPACING_MARK},
{0x011726, 0x011726, PG_U_SPACING_MARK},
{0x011727, 0x01172b, PG_U_NONSPACING_MARK},
- {0x01172c, 0x01172f, PG_U_UNASSIGNED},
{0x011730, 0x011739, PG_U_DECIMAL_NUMBER},
{0x01173a, 0x01173b, PG_U_OTHER_NUMBER},
{0x01173c, 0x01173e, PG_U_OTHER_PUNCTUATION},
{0x01173f, 0x01173f, PG_U_OTHER_SYMBOL},
{0x011740, 0x011746, PG_U_OTHER_LETTER},
- {0x011747, 0x0117ff, PG_U_UNASSIGNED},
{0x011800, 0x01182b, PG_U_OTHER_LETTER},
{0x01182c, 0x01182e, PG_U_SPACING_MARK},
{0x01182f, 0x011837, PG_U_NONSPACING_MARK},
{0x011838, 0x011838, PG_U_SPACING_MARK},
{0x011839, 0x01183a, PG_U_NONSPACING_MARK},
{0x01183b, 0x01183b, PG_U_OTHER_PUNCTUATION},
- {0x01183c, 0x01189f, PG_U_UNASSIGNED},
{0x0118a0, 0x0118bf, PG_U_UPPERCASE_LETTER},
{0x0118c0, 0x0118df, PG_U_LOWERCASE_LETTER},
{0x0118e0, 0x0118e9, PG_U_DECIMAL_NUMBER},
{0x0118ea, 0x0118f2, PG_U_OTHER_NUMBER},
- {0x0118f3, 0x0118fe, PG_U_UNASSIGNED},
{0x0118ff, 0x011906, PG_U_OTHER_LETTER},
- {0x011907, 0x011908, PG_U_UNASSIGNED},
{0x011909, 0x011909, PG_U_OTHER_LETTER},
- {0x01190a, 0x01190b, PG_U_UNASSIGNED},
{0x01190c, 0x011913, PG_U_OTHER_LETTER},
- {0x011914, 0x011914, PG_U_UNASSIGNED},
{0x011915, 0x011916, PG_U_OTHER_LETTER},
- {0x011917, 0x011917, PG_U_UNASSIGNED},
{0x011918, 0x01192f, PG_U_OTHER_LETTER},
{0x011930, 0x011935, PG_U_SPACING_MARK},
- {0x011936, 0x011936, PG_U_UNASSIGNED},
{0x011937, 0x011938, PG_U_SPACING_MARK},
- {0x011939, 0x01193a, PG_U_UNASSIGNED},
{0x01193b, 0x01193c, PG_U_NONSPACING_MARK},
{0x01193d, 0x01193d, PG_U_SPACING_MARK},
{0x01193e, 0x01193e, PG_U_NONSPACING_MARK},
@@ -3406,15 +2923,11 @@ static const pg_category_range unicode_categories[4009] =
{0x011942, 0x011942, PG_U_SPACING_MARK},
{0x011943, 0x011943, PG_U_NONSPACING_MARK},
{0x011944, 0x011946, PG_U_OTHER_PUNCTUATION},
- {0x011947, 0x01194f, PG_U_UNASSIGNED},
{0x011950, 0x011959, PG_U_DECIMAL_NUMBER},
- {0x01195a, 0x01199f, PG_U_UNASSIGNED},
{0x0119a0, 0x0119a7, PG_U_OTHER_LETTER},
- {0x0119a8, 0x0119a9, PG_U_UNASSIGNED},
{0x0119aa, 0x0119d0, PG_U_OTHER_LETTER},
{0x0119d1, 0x0119d3, PG_U_SPACING_MARK},
{0x0119d4, 0x0119d7, PG_U_NONSPACING_MARK},
- {0x0119d8, 0x0119d9, PG_U_UNASSIGNED},
{0x0119da, 0x0119db, PG_U_NONSPACING_MARK},
{0x0119dc, 0x0119df, PG_U_SPACING_MARK},
{0x0119e0, 0x0119e0, PG_U_NONSPACING_MARK},
@@ -3422,7 +2935,6 @@ static const pg_category_range unicode_categories[4009] =
{0x0119e2, 0x0119e2, PG_U_OTHER_PUNCTUATION},
{0x0119e3, 0x0119e3, PG_U_OTHER_LETTER},
{0x0119e4, 0x0119e4, PG_U_SPACING_MARK},
- {0x0119e5, 0x0119ff, PG_U_UNASSIGNED},
{0x011a00, 0x011a00, PG_U_OTHER_LETTER},
{0x011a01, 0x011a0a, PG_U_NONSPACING_MARK},
{0x011a0b, 0x011a32, PG_U_OTHER_LETTER},
@@ -3432,7 +2944,6 @@ static const pg_category_range unicode_categories[4009] =
{0x011a3b, 0x011a3e, PG_U_NONSPACING_MARK},
{0x011a3f, 0x011a46, PG_U_OTHER_PUNCTUATION},
{0x011a47, 0x011a47, PG_U_NONSPACING_MARK},
- {0x011a48, 0x011a4f, PG_U_UNASSIGNED},
{0x011a50, 0x011a50, PG_U_OTHER_LETTER},
{0x011a51, 0x011a56, PG_U_NONSPACING_MARK},
{0x011a57, 0x011a58, PG_U_SPACING_MARK},
@@ -3444,136 +2955,93 @@ static const pg_category_range unicode_categories[4009] =
{0x011a9a, 0x011a9c, PG_U_OTHER_PUNCTUATION},
{0x011a9d, 0x011a9d, PG_U_OTHER_LETTER},
{0x011a9e, 0x011aa2, PG_U_OTHER_PUNCTUATION},
- {0x011aa3, 0x011aaf, PG_U_UNASSIGNED},
{0x011ab0, 0x011af8, PG_U_OTHER_LETTER},
- {0x011af9, 0x011aff, PG_U_UNASSIGNED},
{0x011b00, 0x011b09, PG_U_OTHER_PUNCTUATION},
- {0x011b0a, 0x011bff, PG_U_UNASSIGNED},
{0x011c00, 0x011c08, PG_U_OTHER_LETTER},
- {0x011c09, 0x011c09, PG_U_UNASSIGNED},
{0x011c0a, 0x011c2e, PG_U_OTHER_LETTER},
{0x011c2f, 0x011c2f, PG_U_SPACING_MARK},
{0x011c30, 0x011c36, PG_U_NONSPACING_MARK},
- {0x011c37, 0x011c37, PG_U_UNASSIGNED},
{0x011c38, 0x011c3d, PG_U_NONSPACING_MARK},
{0x011c3e, 0x011c3e, PG_U_SPACING_MARK},
{0x011c3f, 0x011c3f, PG_U_NONSPACING_MARK},
{0x011c40, 0x011c40, PG_U_OTHER_LETTER},
{0x011c41, 0x011c45, PG_U_OTHER_PUNCTUATION},
- {0x011c46, 0x011c4f, PG_U_UNASSIGNED},
{0x011c50, 0x011c59, PG_U_DECIMAL_NUMBER},
{0x011c5a, 0x011c6c, PG_U_OTHER_NUMBER},
- {0x011c6d, 0x011c6f, PG_U_UNASSIGNED},
{0x011c70, 0x011c71, PG_U_OTHER_PUNCTUATION},
{0x011c72, 0x011c8f, PG_U_OTHER_LETTER},
- {0x011c90, 0x011c91, PG_U_UNASSIGNED},
{0x011c92, 0x011ca7, PG_U_NONSPACING_MARK},
- {0x011ca8, 0x011ca8, PG_U_UNASSIGNED},
{0x011ca9, 0x011ca9, PG_U_SPACING_MARK},
{0x011caa, 0x011cb0, PG_U_NONSPACING_MARK},
{0x011cb1, 0x011cb1, PG_U_SPACING_MARK},
{0x011cb2, 0x011cb3, PG_U_NONSPACING_MARK},
{0x011cb4, 0x011cb4, PG_U_SPACING_MARK},
{0x011cb5, 0x011cb6, PG_U_NONSPACING_MARK},
- {0x011cb7, 0x011cff, PG_U_UNASSIGNED},
{0x011d00, 0x011d06, PG_U_OTHER_LETTER},
- {0x011d07, 0x011d07, PG_U_UNASSIGNED},
{0x011d08, 0x011d09, PG_U_OTHER_LETTER},
- {0x011d0a, 0x011d0a, PG_U_UNASSIGNED},
{0x011d0b, 0x011d30, PG_U_OTHER_LETTER},
{0x011d31, 0x011d36, PG_U_NONSPACING_MARK},
- {0x011d37, 0x011d39, PG_U_UNASSIGNED},
{0x011d3a, 0x011d3a, PG_U_NONSPACING_MARK},
- {0x011d3b, 0x011d3b, PG_U_UNASSIGNED},
{0x011d3c, 0x011d3d, PG_U_NONSPACING_MARK},
- {0x011d3e, 0x011d3e, PG_U_UNASSIGNED},
{0x011d3f, 0x011d45, PG_U_NONSPACING_MARK},
{0x011d46, 0x011d46, PG_U_OTHER_LETTER},
{0x011d47, 0x011d47, PG_U_NONSPACING_MARK},
- {0x011d48, 0x011d4f, PG_U_UNASSIGNED},
{0x011d50, 0x011d59, PG_U_DECIMAL_NUMBER},
- {0x011d5a, 0x011d5f, PG_U_UNASSIGNED},
{0x011d60, 0x011d65, PG_U_OTHER_LETTER},
- {0x011d66, 0x011d66, PG_U_UNASSIGNED},
{0x011d67, 0x011d68, PG_U_OTHER_LETTER},
- {0x011d69, 0x011d69, PG_U_UNASSIGNED},
{0x011d6a, 0x011d89, PG_U_OTHER_LETTER},
{0x011d8a, 0x011d8e, PG_U_SPACING_MARK},
- {0x011d8f, 0x011d8f, PG_U_UNASSIGNED},
{0x011d90, 0x011d91, PG_U_NONSPACING_MARK},
- {0x011d92, 0x011d92, PG_U_UNASSIGNED},
{0x011d93, 0x011d94, PG_U_SPACING_MARK},
{0x011d95, 0x011d95, PG_U_NONSPACING_MARK},
{0x011d96, 0x011d96, PG_U_SPACING_MARK},
{0x011d97, 0x011d97, PG_U_NONSPACING_MARK},
{0x011d98, 0x011d98, PG_U_OTHER_LETTER},
- {0x011d99, 0x011d9f, PG_U_UNASSIGNED},
{0x011da0, 0x011da9, PG_U_DECIMAL_NUMBER},
- {0x011daa, 0x011edf, PG_U_UNASSIGNED},
{0x011ee0, 0x011ef2, PG_U_OTHER_LETTER},
{0x011ef3, 0x011ef4, PG_U_NONSPACING_MARK},
{0x011ef5, 0x011ef6, PG_U_SPACING_MARK},
{0x011ef7, 0x011ef8, PG_U_OTHER_PUNCTUATION},
- {0x011ef9, 0x011eff, PG_U_UNASSIGNED},
{0x011f00, 0x011f01, PG_U_NONSPACING_MARK},
{0x011f02, 0x011f02, PG_U_OTHER_LETTER},
{0x011f03, 0x011f03, PG_U_SPACING_MARK},
{0x011f04, 0x011f10, PG_U_OTHER_LETTER},
- {0x011f11, 0x011f11, PG_U_UNASSIGNED},
{0x011f12, 0x011f33, PG_U_OTHER_LETTER},
{0x011f34, 0x011f35, PG_U_SPACING_MARK},
{0x011f36, 0x011f3a, PG_U_NONSPACING_MARK},
- {0x011f3b, 0x011f3d, PG_U_UNASSIGNED},
{0x011f3e, 0x011f3f, PG_U_SPACING_MARK},
{0x011f40, 0x011f40, PG_U_NONSPACING_MARK},
{0x011f41, 0x011f41, PG_U_SPACING_MARK},
{0x011f42, 0x011f42, PG_U_NONSPACING_MARK},
{0x011f43, 0x011f4f, PG_U_OTHER_PUNCTUATION},
{0x011f50, 0x011f59, PG_U_DECIMAL_NUMBER},
- {0x011f5a, 0x011faf, PG_U_UNASSIGNED},
{0x011fb0, 0x011fb0, PG_U_OTHER_LETTER},
- {0x011fb1, 0x011fbf, PG_U_UNASSIGNED},
{0x011fc0, 0x011fd4, PG_U_OTHER_NUMBER},
{0x011fd5, 0x011fdc, PG_U_OTHER_SYMBOL},
{0x011fdd, 0x011fe0, PG_U_CURRENCY_SYMBOL},
{0x011fe1, 0x011ff1, PG_U_OTHER_SYMBOL},
- {0x011ff2, 0x011ffe, PG_U_UNASSIGNED},
{0x011fff, 0x011fff, PG_U_OTHER_PUNCTUATION},
{0x012000, 0x012399, PG_U_OTHER_LETTER},
- {0x01239a, 0x0123ff, PG_U_UNASSIGNED},
{0x012400, 0x01246e, PG_U_LETTER_NUMBER},
- {0x01246f, 0x01246f, PG_U_UNASSIGNED},
{0x012470, 0x012474, PG_U_OTHER_PUNCTUATION},
- {0x012475, 0x01247f, PG_U_UNASSIGNED},
{0x012480, 0x012543, PG_U_OTHER_LETTER},
- {0x012544, 0x012f8f, PG_U_UNASSIGNED},
{0x012f90, 0x012ff0, PG_U_OTHER_LETTER},
{0x012ff1, 0x012ff2, PG_U_OTHER_PUNCTUATION},
- {0x012ff3, 0x012fff, PG_U_UNASSIGNED},
{0x013000, 0x01342f, PG_U_OTHER_LETTER},
{0x013430, 0x01343f, PG_U_FORMAT},
{0x013440, 0x013440, PG_U_NONSPACING_MARK},
{0x013441, 0x013446, PG_U_OTHER_LETTER},
{0x013447, 0x013455, PG_U_NONSPACING_MARK},
- {0x013456, 0x0143ff, PG_U_UNASSIGNED},
{0x014400, 0x014646, PG_U_OTHER_LETTER},
- {0x014647, 0x0167ff, PG_U_UNASSIGNED},
{0x016800, 0x016a38, PG_U_OTHER_LETTER},
- {0x016a39, 0x016a3f, PG_U_UNASSIGNED},
{0x016a40, 0x016a5e, PG_U_OTHER_LETTER},
- {0x016a5f, 0x016a5f, PG_U_UNASSIGNED},
{0x016a60, 0x016a69, PG_U_DECIMAL_NUMBER},
- {0x016a6a, 0x016a6d, PG_U_UNASSIGNED},
{0x016a6e, 0x016a6f, PG_U_OTHER_PUNCTUATION},
{0x016a70, 0x016abe, PG_U_OTHER_LETTER},
- {0x016abf, 0x016abf, PG_U_UNASSIGNED},
{0x016ac0, 0x016ac9, PG_U_DECIMAL_NUMBER},
- {0x016aca, 0x016acf, PG_U_UNASSIGNED},
{0x016ad0, 0x016aed, PG_U_OTHER_LETTER},
- {0x016aee, 0x016aef, PG_U_UNASSIGNED},
{0x016af0, 0x016af4, PG_U_NONSPACING_MARK},
{0x016af5, 0x016af5, PG_U_OTHER_PUNCTUATION},
- {0x016af6, 0x016aff, PG_U_UNASSIGNED},
{0x016b00, 0x016b2f, PG_U_OTHER_LETTER},
{0x016b30, 0x016b36, PG_U_NONSPACING_MARK},
{0x016b37, 0x016b3b, PG_U_OTHER_PUNCTUATION},
@@ -3581,83 +3049,50 @@ static const pg_category_range unicode_categories[4009] =
{0x016b40, 0x016b43, PG_U_MODIFIER_LETTER},
{0x016b44, 0x016b44, PG_U_OTHER_PUNCTUATION},
{0x016b45, 0x016b45, PG_U_OTHER_SYMBOL},
- {0x016b46, 0x016b4f, PG_U_UNASSIGNED},
{0x016b50, 0x016b59, PG_U_DECIMAL_NUMBER},
- {0x016b5a, 0x016b5a, PG_U_UNASSIGNED},
{0x016b5b, 0x016b61, PG_U_OTHER_NUMBER},
- {0x016b62, 0x016b62, PG_U_UNASSIGNED},
{0x016b63, 0x016b77, PG_U_OTHER_LETTER},
- {0x016b78, 0x016b7c, PG_U_UNASSIGNED},
{0x016b7d, 0x016b8f, PG_U_OTHER_LETTER},
- {0x016b90, 0x016e3f, PG_U_UNASSIGNED},
{0x016e40, 0x016e5f, PG_U_UPPERCASE_LETTER},
{0x016e60, 0x016e7f, PG_U_LOWERCASE_LETTER},
{0x016e80, 0x016e96, PG_U_OTHER_NUMBER},
{0x016e97, 0x016e9a, PG_U_OTHER_PUNCTUATION},
- {0x016e9b, 0x016eff, PG_U_UNASSIGNED},
{0x016f00, 0x016f4a, PG_U_OTHER_LETTER},
- {0x016f4b, 0x016f4e, PG_U_UNASSIGNED},
{0x016f4f, 0x016f4f, PG_U_NONSPACING_MARK},
{0x016f50, 0x016f50, PG_U_OTHER_LETTER},
{0x016f51, 0x016f87, PG_U_SPACING_MARK},
- {0x016f88, 0x016f8e, PG_U_UNASSIGNED},
{0x016f8f, 0x016f92, PG_U_NONSPACING_MARK},
{0x016f93, 0x016f9f, PG_U_MODIFIER_LETTER},
- {0x016fa0, 0x016fdf, PG_U_UNASSIGNED},
{0x016fe0, 0x016fe1, PG_U_MODIFIER_LETTER},
{0x016fe2, 0x016fe2, PG_U_OTHER_PUNCTUATION},
{0x016fe3, 0x016fe3, PG_U_MODIFIER_LETTER},
{0x016fe4, 0x016fe4, PG_U_NONSPACING_MARK},
- {0x016fe5, 0x016fef, PG_U_UNASSIGNED},
{0x016ff0, 0x016ff1, PG_U_SPACING_MARK},
- {0x016ff2, 0x016fff, PG_U_UNASSIGNED},
{0x017000, 0x0187f7, PG_U_OTHER_LETTER},
- {0x0187f8, 0x0187ff, PG_U_UNASSIGNED},
{0x018800, 0x018cd5, PG_U_OTHER_LETTER},
- {0x018cd6, 0x018cff, PG_U_UNASSIGNED},
{0x018d00, 0x018d08, PG_U_OTHER_LETTER},
- {0x018d09, 0x01afef, PG_U_UNASSIGNED},
{0x01aff0, 0x01aff3, PG_U_MODIFIER_LETTER},
- {0x01aff4, 0x01aff4, PG_U_UNASSIGNED},
{0x01aff5, 0x01affb, PG_U_MODIFIER_LETTER},
- {0x01affc, 0x01affc, PG_U_UNASSIGNED},
{0x01affd, 0x01affe, PG_U_MODIFIER_LETTER},
- {0x01afff, 0x01afff, PG_U_UNASSIGNED},
{0x01b000, 0x01b122, PG_U_OTHER_LETTER},
- {0x01b123, 0x01b131, PG_U_UNASSIGNED},
{0x01b132, 0x01b132, PG_U_OTHER_LETTER},
- {0x01b133, 0x01b14f, PG_U_UNASSIGNED},
{0x01b150, 0x01b152, PG_U_OTHER_LETTER},
- {0x01b153, 0x01b154, PG_U_UNASSIGNED},
{0x01b155, 0x01b155, PG_U_OTHER_LETTER},
- {0x01b156, 0x01b163, PG_U_UNASSIGNED},
{0x01b164, 0x01b167, PG_U_OTHER_LETTER},
- {0x01b168, 0x01b16f, PG_U_UNASSIGNED},
{0x01b170, 0x01b2fb, PG_U_OTHER_LETTER},
- {0x01b2fc, 0x01bbff, PG_U_UNASSIGNED},
{0x01bc00, 0x01bc6a, PG_U_OTHER_LETTER},
- {0x01bc6b, 0x01bc6f, PG_U_UNASSIGNED},
{0x01bc70, 0x01bc7c, PG_U_OTHER_LETTER},
- {0x01bc7d, 0x01bc7f, PG_U_UNASSIGNED},
{0x01bc80, 0x01bc88, PG_U_OTHER_LETTER},
- {0x01bc89, 0x01bc8f, PG_U_UNASSIGNED},
{0x01bc90, 0x01bc99, PG_U_OTHER_LETTER},
- {0x01bc9a, 0x01bc9b, PG_U_UNASSIGNED},
{0x01bc9c, 0x01bc9c, PG_U_OTHER_SYMBOL},
{0x01bc9d, 0x01bc9e, PG_U_NONSPACING_MARK},
{0x01bc9f, 0x01bc9f, PG_U_OTHER_PUNCTUATION},
{0x01bca0, 0x01bca3, PG_U_FORMAT},
- {0x01bca4, 0x01ceff, PG_U_UNASSIGNED},
{0x01cf00, 0x01cf2d, PG_U_NONSPACING_MARK},
- {0x01cf2e, 0x01cf2f, PG_U_UNASSIGNED},
{0x01cf30, 0x01cf46, PG_U_NONSPACING_MARK},
- {0x01cf47, 0x01cf4f, PG_U_UNASSIGNED},
{0x01cf50, 0x01cfc3, PG_U_OTHER_SYMBOL},
- {0x01cfc4, 0x01cfff, PG_U_UNASSIGNED},
{0x01d000, 0x01d0f5, PG_U_OTHER_SYMBOL},
- {0x01d0f6, 0x01d0ff, PG_U_UNASSIGNED},
{0x01d100, 0x01d126, PG_U_OTHER_SYMBOL},
- {0x01d127, 0x01d128, PG_U_UNASSIGNED},
{0x01d129, 0x01d164, PG_U_OTHER_SYMBOL},
{0x01d165, 0x01d166, PG_U_SPACING_MARK},
{0x01d167, 0x01d169, PG_U_NONSPACING_MARK},
@@ -3670,66 +3105,42 @@ static const pg_category_range unicode_categories[4009] =
{0x01d18c, 0x01d1a9, PG_U_OTHER_SYMBOL},
{0x01d1aa, 0x01d1ad, PG_U_NONSPACING_MARK},
{0x01d1ae, 0x01d1ea, PG_U_OTHER_SYMBOL},
- {0x01d1eb, 0x01d1ff, PG_U_UNASSIGNED},
{0x01d200, 0x01d241, PG_U_OTHER_SYMBOL},
{0x01d242, 0x01d244, PG_U_NONSPACING_MARK},
{0x01d245, 0x01d245, PG_U_OTHER_SYMBOL},
- {0x01d246, 0x01d2bf, PG_U_UNASSIGNED},
{0x01d2c0, 0x01d2d3, PG_U_OTHER_NUMBER},
- {0x01d2d4, 0x01d2df, PG_U_UNASSIGNED},
{0x01d2e0, 0x01d2f3, PG_U_OTHER_NUMBER},
- {0x01d2f4, 0x01d2ff, PG_U_UNASSIGNED},
{0x01d300, 0x01d356, PG_U_OTHER_SYMBOL},
- {0x01d357, 0x01d35f, PG_U_UNASSIGNED},
{0x01d360, 0x01d378, PG_U_OTHER_NUMBER},
- {0x01d379, 0x01d3ff, PG_U_UNASSIGNED},
{0x01d400, 0x01d419, PG_U_UPPERCASE_LETTER},
{0x01d41a, 0x01d433, PG_U_LOWERCASE_LETTER},
{0x01d434, 0x01d44d, PG_U_UPPERCASE_LETTER},
{0x01d44e, 0x01d454, PG_U_LOWERCASE_LETTER},
- {0x01d455, 0x01d455, PG_U_UNASSIGNED},
{0x01d456, 0x01d467, PG_U_LOWERCASE_LETTER},
{0x01d468, 0x01d481, PG_U_UPPERCASE_LETTER},
{0x01d482, 0x01d49b, PG_U_LOWERCASE_LETTER},
{0x01d49c, 0x01d49c, PG_U_UPPERCASE_LETTER},
- {0x01d49d, 0x01d49d, PG_U_UNASSIGNED},
{0x01d49e, 0x01d49f, PG_U_UPPERCASE_LETTER},
- {0x01d4a0, 0x01d4a1, PG_U_UNASSIGNED},
{0x01d4a2, 0x01d4a2, PG_U_UPPERCASE_LETTER},
- {0x01d4a3, 0x01d4a4, PG_U_UNASSIGNED},
{0x01d4a5, 0x01d4a6, PG_U_UPPERCASE_LETTER},
- {0x01d4a7, 0x01d4a8, PG_U_UNASSIGNED},
{0x01d4a9, 0x01d4ac, PG_U_UPPERCASE_LETTER},
- {0x01d4ad, 0x01d4ad, PG_U_UNASSIGNED},
{0x01d4ae, 0x01d4b5, PG_U_UPPERCASE_LETTER},
{0x01d4b6, 0x01d4b9, PG_U_LOWERCASE_LETTER},
- {0x01d4ba, 0x01d4ba, PG_U_UNASSIGNED},
{0x01d4bb, 0x01d4bb, PG_U_LOWERCASE_LETTER},
- {0x01d4bc, 0x01d4bc, PG_U_UNASSIGNED},
{0x01d4bd, 0x01d4c3, PG_U_LOWERCASE_LETTER},
- {0x01d4c4, 0x01d4c4, PG_U_UNASSIGNED},
{0x01d4c5, 0x01d4cf, PG_U_LOWERCASE_LETTER},
{0x01d4d0, 0x01d4e9, PG_U_UPPERCASE_LETTER},
{0x01d4ea, 0x01d503, PG_U_LOWERCASE_LETTER},
{0x01d504, 0x01d505, PG_U_UPPERCASE_LETTER},
- {0x01d506, 0x01d506, PG_U_UNASSIGNED},
{0x01d507, 0x01d50a, PG_U_UPPERCASE_LETTER},
- {0x01d50b, 0x01d50c, PG_U_UNASSIGNED},
{0x01d50d, 0x01d514, PG_U_UPPERCASE_LETTER},
- {0x01d515, 0x01d515, PG_U_UNASSIGNED},
{0x01d516, 0x01d51c, PG_U_UPPERCASE_LETTER},
- {0x01d51d, 0x01d51d, PG_U_UNASSIGNED},
{0x01d51e, 0x01d537, PG_U_LOWERCASE_LETTER},
{0x01d538, 0x01d539, PG_U_UPPERCASE_LETTER},
- {0x01d53a, 0x01d53a, PG_U_UNASSIGNED},
{0x01d53b, 0x01d53e, PG_U_UPPERCASE_LETTER},
- {0x01d53f, 0x01d53f, PG_U_UNASSIGNED},
{0x01d540, 0x01d544, PG_U_UPPERCASE_LETTER},
- {0x01d545, 0x01d545, PG_U_UNASSIGNED},
{0x01d546, 0x01d546, PG_U_UPPERCASE_LETTER},
- {0x01d547, 0x01d549, PG_U_UNASSIGNED},
{0x01d54a, 0x01d550, PG_U_UPPERCASE_LETTER},
- {0x01d551, 0x01d551, PG_U_UNASSIGNED},
{0x01d552, 0x01d56b, PG_U_LOWERCASE_LETTER},
{0x01d56c, 0x01d585, PG_U_UPPERCASE_LETTER},
{0x01d586, 0x01d59f, PG_U_LOWERCASE_LETTER},
@@ -3743,7 +3154,6 @@ static const pg_category_range unicode_categories[4009] =
{0x01d656, 0x01d66f, PG_U_LOWERCASE_LETTER},
{0x01d670, 0x01d689, PG_U_UPPERCASE_LETTER},
{0x01d68a, 0x01d6a5, PG_U_LOWERCASE_LETTER},
- {0x01d6a6, 0x01d6a7, PG_U_UNASSIGNED},
{0x01d6a8, 0x01d6c0, PG_U_UPPERCASE_LETTER},
{0x01d6c1, 0x01d6c1, PG_U_MATH_SYMBOL},
{0x01d6c2, 0x01d6da, PG_U_LOWERCASE_LETTER},
@@ -3771,7 +3181,6 @@ static const pg_category_range unicode_categories[4009] =
{0x01d7c4, 0x01d7c9, PG_U_LOWERCASE_LETTER},
{0x01d7ca, 0x01d7ca, PG_U_UPPERCASE_LETTER},
{0x01d7cb, 0x01d7cb, PG_U_LOWERCASE_LETTER},
- {0x01d7cc, 0x01d7cd, PG_U_UNASSIGNED},
{0x01d7ce, 0x01d7ff, PG_U_DECIMAL_NUMBER},
{0x01d800, 0x01d9ff, PG_U_OTHER_SYMBOL},
{0x01da00, 0x01da36, PG_U_NONSPACING_MARK},
@@ -3783,258 +3192,142 @@ static const pg_category_range unicode_categories[4009] =
{0x01da84, 0x01da84, PG_U_NONSPACING_MARK},
{0x01da85, 0x01da86, PG_U_OTHER_SYMBOL},
{0x01da87, 0x01da8b, PG_U_OTHER_PUNCTUATION},
- {0x01da8c, 0x01da9a, PG_U_UNASSIGNED},
{0x01da9b, 0x01da9f, PG_U_NONSPACING_MARK},
- {0x01daa0, 0x01daa0, PG_U_UNASSIGNED},
{0x01daa1, 0x01daaf, PG_U_NONSPACING_MARK},
- {0x01dab0, 0x01deff, PG_U_UNASSIGNED},
{0x01df00, 0x01df09, PG_U_LOWERCASE_LETTER},
{0x01df0a, 0x01df0a, PG_U_OTHER_LETTER},
{0x01df0b, 0x01df1e, PG_U_LOWERCASE_LETTER},
- {0x01df1f, 0x01df24, PG_U_UNASSIGNED},
{0x01df25, 0x01df2a, PG_U_LOWERCASE_LETTER},
- {0x01df2b, 0x01dfff, PG_U_UNASSIGNED},
{0x01e000, 0x01e006, PG_U_NONSPACING_MARK},
- {0x01e007, 0x01e007, PG_U_UNASSIGNED},
{0x01e008, 0x01e018, PG_U_NONSPACING_MARK},
- {0x01e019, 0x01e01a, PG_U_UNASSIGNED},
{0x01e01b, 0x01e021, PG_U_NONSPACING_MARK},
- {0x01e022, 0x01e022, PG_U_UNASSIGNED},
{0x01e023, 0x01e024, PG_U_NONSPACING_MARK},
- {0x01e025, 0x01e025, PG_U_UNASSIGNED},
{0x01e026, 0x01e02a, PG_U_NONSPACING_MARK},
- {0x01e02b, 0x01e02f, PG_U_UNASSIGNED},
{0x01e030, 0x01e06d, PG_U_MODIFIER_LETTER},
- {0x01e06e, 0x01e08e, PG_U_UNASSIGNED},
{0x01e08f, 0x01e08f, PG_U_NONSPACING_MARK},
- {0x01e090, 0x01e0ff, PG_U_UNASSIGNED},
{0x01e100, 0x01e12c, PG_U_OTHER_LETTER},
- {0x01e12d, 0x01e12f, PG_U_UNASSIGNED},
{0x01e130, 0x01e136, PG_U_NONSPACING_MARK},
{0x01e137, 0x01e13d, PG_U_MODIFIER_LETTER},
- {0x01e13e, 0x01e13f, PG_U_UNASSIGNED},
{0x01e140, 0x01e149, PG_U_DECIMAL_NUMBER},
- {0x01e14a, 0x01e14d, PG_U_UNASSIGNED},
{0x01e14e, 0x01e14e, PG_U_OTHER_LETTER},
{0x01e14f, 0x01e14f, PG_U_OTHER_SYMBOL},
- {0x01e150, 0x01e28f, PG_U_UNASSIGNED},
{0x01e290, 0x01e2ad, PG_U_OTHER_LETTER},
{0x01e2ae, 0x01e2ae, PG_U_NONSPACING_MARK},
- {0x01e2af, 0x01e2bf, PG_U_UNASSIGNED},
{0x01e2c0, 0x01e2eb, PG_U_OTHER_LETTER},
{0x01e2ec, 0x01e2ef, PG_U_NONSPACING_MARK},
{0x01e2f0, 0x01e2f9, PG_U_DECIMAL_NUMBER},
- {0x01e2fa, 0x01e2fe, PG_U_UNASSIGNED},
{0x01e2ff, 0x01e2ff, PG_U_CURRENCY_SYMBOL},
- {0x01e300, 0x01e4cf, PG_U_UNASSIGNED},
{0x01e4d0, 0x01e4ea, PG_U_OTHER_LETTER},
{0x01e4eb, 0x01e4eb, PG_U_MODIFIER_LETTER},
{0x01e4ec, 0x01e4ef, PG_U_NONSPACING_MARK},
{0x01e4f0, 0x01e4f9, PG_U_DECIMAL_NUMBER},
- {0x01e4fa, 0x01e7df, PG_U_UNASSIGNED},
{0x01e7e0, 0x01e7e6, PG_U_OTHER_LETTER},
- {0x01e7e7, 0x01e7e7, PG_U_UNASSIGNED},
{0x01e7e8, 0x01e7eb, PG_U_OTHER_LETTER},
- {0x01e7ec, 0x01e7ec, PG_U_UNASSIGNED},
{0x01e7ed, 0x01e7ee, PG_U_OTHER_LETTER},
- {0x01e7ef, 0x01e7ef, PG_U_UNASSIGNED},
{0x01e7f0, 0x01e7fe, PG_U_OTHER_LETTER},
- {0x01e7ff, 0x01e7ff, PG_U_UNASSIGNED},
{0x01e800, 0x01e8c4, PG_U_OTHER_LETTER},
- {0x01e8c5, 0x01e8c6, PG_U_UNASSIGNED},
{0x01e8c7, 0x01e8cf, PG_U_OTHER_NUMBER},
{0x01e8d0, 0x01e8d6, PG_U_NONSPACING_MARK},
- {0x01e8d7, 0x01e8ff, PG_U_UNASSIGNED},
{0x01e900, 0x01e921, PG_U_UPPERCASE_LETTER},
{0x01e922, 0x01e943, PG_U_LOWERCASE_LETTER},
{0x01e944, 0x01e94a, PG_U_NONSPACING_MARK},
{0x01e94b, 0x01e94b, PG_U_MODIFIER_LETTER},
- {0x01e94c, 0x01e94f, PG_U_UNASSIGNED},
{0x01e950, 0x01e959, PG_U_DECIMAL_NUMBER},
- {0x01e95a, 0x01e95d, PG_U_UNASSIGNED},
{0x01e95e, 0x01e95f, PG_U_OTHER_PUNCTUATION},
- {0x01e960, 0x01ec70, PG_U_UNASSIGNED},
{0x01ec71, 0x01ecab, PG_U_OTHER_NUMBER},
{0x01ecac, 0x01ecac, PG_U_OTHER_SYMBOL},
{0x01ecad, 0x01ecaf, PG_U_OTHER_NUMBER},
{0x01ecb0, 0x01ecb0, PG_U_CURRENCY_SYMBOL},
{0x01ecb1, 0x01ecb4, PG_U_OTHER_NUMBER},
- {0x01ecb5, 0x01ed00, PG_U_UNASSIGNED},
{0x01ed01, 0x01ed2d, PG_U_OTHER_NUMBER},
{0x01ed2e, 0x01ed2e, PG_U_OTHER_SYMBOL},
{0x01ed2f, 0x01ed3d, PG_U_OTHER_NUMBER},
- {0x01ed3e, 0x01edff, PG_U_UNASSIGNED},
{0x01ee00, 0x01ee03, PG_U_OTHER_LETTER},
- {0x01ee04, 0x01ee04, PG_U_UNASSIGNED},
{0x01ee05, 0x01ee1f, PG_U_OTHER_LETTER},
- {0x01ee20, 0x01ee20, PG_U_UNASSIGNED},
{0x01ee21, 0x01ee22, PG_U_OTHER_LETTER},
- {0x01ee23, 0x01ee23, PG_U_UNASSIGNED},
{0x01ee24, 0x01ee24, PG_U_OTHER_LETTER},
- {0x01ee25, 0x01ee26, PG_U_UNASSIGNED},
{0x01ee27, 0x01ee27, PG_U_OTHER_LETTER},
- {0x01ee28, 0x01ee28, PG_U_UNASSIGNED},
{0x01ee29, 0x01ee32, PG_U_OTHER_LETTER},
- {0x01ee33, 0x01ee33, PG_U_UNASSIGNED},
{0x01ee34, 0x01ee37, PG_U_OTHER_LETTER},
- {0x01ee38, 0x01ee38, PG_U_UNASSIGNED},
{0x01ee39, 0x01ee39, PG_U_OTHER_LETTER},
- {0x01ee3a, 0x01ee3a, PG_U_UNASSIGNED},
{0x01ee3b, 0x01ee3b, PG_U_OTHER_LETTER},
- {0x01ee3c, 0x01ee41, PG_U_UNASSIGNED},
{0x01ee42, 0x01ee42, PG_U_OTHER_LETTER},
- {0x01ee43, 0x01ee46, PG_U_UNASSIGNED},
{0x01ee47, 0x01ee47, PG_U_OTHER_LETTER},
- {0x01ee48, 0x01ee48, PG_U_UNASSIGNED},
{0x01ee49, 0x01ee49, PG_U_OTHER_LETTER},
- {0x01ee4a, 0x01ee4a, PG_U_UNASSIGNED},
{0x01ee4b, 0x01ee4b, PG_U_OTHER_LETTER},
- {0x01ee4c, 0x01ee4c, PG_U_UNASSIGNED},
{0x01ee4d, 0x01ee4f, PG_U_OTHER_LETTER},
- {0x01ee50, 0x01ee50, PG_U_UNASSIGNED},
{0x01ee51, 0x01ee52, PG_U_OTHER_LETTER},
- {0x01ee53, 0x01ee53, PG_U_UNASSIGNED},
{0x01ee54, 0x01ee54, PG_U_OTHER_LETTER},
- {0x01ee55, 0x01ee56, PG_U_UNASSIGNED},
{0x01ee57, 0x01ee57, PG_U_OTHER_LETTER},
- {0x01ee58, 0x01ee58, PG_U_UNASSIGNED},
{0x01ee59, 0x01ee59, PG_U_OTHER_LETTER},
- {0x01ee5a, 0x01ee5a, PG_U_UNASSIGNED},
{0x01ee5b, 0x01ee5b, PG_U_OTHER_LETTER},
- {0x01ee5c, 0x01ee5c, PG_U_UNASSIGNED},
{0x01ee5d, 0x01ee5d, PG_U_OTHER_LETTER},
- {0x01ee5e, 0x01ee5e, PG_U_UNASSIGNED},
{0x01ee5f, 0x01ee5f, PG_U_OTHER_LETTER},
- {0x01ee60, 0x01ee60, PG_U_UNASSIGNED},
{0x01ee61, 0x01ee62, PG_U_OTHER_LETTER},
- {0x01ee63, 0x01ee63, PG_U_UNASSIGNED},
{0x01ee64, 0x01ee64, PG_U_OTHER_LETTER},
- {0x01ee65, 0x01ee66, PG_U_UNASSIGNED},
{0x01ee67, 0x01ee6a, PG_U_OTHER_LETTER},
- {0x01ee6b, 0x01ee6b, PG_U_UNASSIGNED},
{0x01ee6c, 0x01ee72, PG_U_OTHER_LETTER},
- {0x01ee73, 0x01ee73, PG_U_UNASSIGNED},
{0x01ee74, 0x01ee77, PG_U_OTHER_LETTER},
- {0x01ee78, 0x01ee78, PG_U_UNASSIGNED},
{0x01ee79, 0x01ee7c, PG_U_OTHER_LETTER},
- {0x01ee7d, 0x01ee7d, PG_U_UNASSIGNED},
{0x01ee7e, 0x01ee7e, PG_U_OTHER_LETTER},
- {0x01ee7f, 0x01ee7f, PG_U_UNASSIGNED},
{0x01ee80, 0x01ee89, PG_U_OTHER_LETTER},
- {0x01ee8a, 0x01ee8a, PG_U_UNASSIGNED},
{0x01ee8b, 0x01ee9b, PG_U_OTHER_LETTER},
- {0x01ee9c, 0x01eea0, PG_U_UNASSIGNED},
{0x01eea1, 0x01eea3, PG_U_OTHER_LETTER},
- {0x01eea4, 0x01eea4, PG_U_UNASSIGNED},
{0x01eea5, 0x01eea9, PG_U_OTHER_LETTER},
- {0x01eeaa, 0x01eeaa, PG_U_UNASSIGNED},
{0x01eeab, 0x01eebb, PG_U_OTHER_LETTER},
- {0x01eebc, 0x01eeef, PG_U_UNASSIGNED},
{0x01eef0, 0x01eef1, PG_U_MATH_SYMBOL},
- {0x01eef2, 0x01efff, PG_U_UNASSIGNED},
{0x01f000, 0x01f02b, PG_U_OTHER_SYMBOL},
- {0x01f02c, 0x01f02f, PG_U_UNASSIGNED},
{0x01f030, 0x01f093, PG_U_OTHER_SYMBOL},
- {0x01f094, 0x01f09f, PG_U_UNASSIGNED},
{0x01f0a0, 0x01f0ae, PG_U_OTHER_SYMBOL},
- {0x01f0af, 0x01f0b0, PG_U_UNASSIGNED},
{0x01f0b1, 0x01f0bf, PG_U_OTHER_SYMBOL},
- {0x01f0c0, 0x01f0c0, PG_U_UNASSIGNED},
{0x01f0c1, 0x01f0cf, PG_U_OTHER_SYMBOL},
- {0x01f0d0, 0x01f0d0, PG_U_UNASSIGNED},
{0x01f0d1, 0x01f0f5, PG_U_OTHER_SYMBOL},
- {0x01f0f6, 0x01f0ff, PG_U_UNASSIGNED},
{0x01f100, 0x01f10c, PG_U_OTHER_NUMBER},
{0x01f10d, 0x01f1ad, PG_U_OTHER_SYMBOL},
- {0x01f1ae, 0x01f1e5, PG_U_UNASSIGNED},
{0x01f1e6, 0x01f202, PG_U_OTHER_SYMBOL},
- {0x01f203, 0x01f20f, PG_U_UNASSIGNED},
{0x01f210, 0x01f23b, PG_U_OTHER_SYMBOL},
- {0x01f23c, 0x01f23f, PG_U_UNASSIGNED},
{0x01f240, 0x01f248, PG_U_OTHER_SYMBOL},
- {0x01f249, 0x01f24f, PG_U_UNASSIGNED},
{0x01f250, 0x01f251, PG_U_OTHER_SYMBOL},
- {0x01f252, 0x01f25f, PG_U_UNASSIGNED},
{0x01f260, 0x01f265, PG_U_OTHER_SYMBOL},
- {0x01f266, 0x01f2ff, PG_U_UNASSIGNED},
{0x01f300, 0x01f3fa, PG_U_OTHER_SYMBOL},
{0x01f3fb, 0x01f3ff, PG_U_MODIFIER_SYMBOL},
{0x01f400, 0x01f6d7, PG_U_OTHER_SYMBOL},
- {0x01f6d8, 0x01f6db, PG_U_UNASSIGNED},
{0x01f6dc, 0x01f6ec, PG_U_OTHER_SYMBOL},
- {0x01f6ed, 0x01f6ef, PG_U_UNASSIGNED},
{0x01f6f0, 0x01f6fc, PG_U_OTHER_SYMBOL},
- {0x01f6fd, 0x01f6ff, PG_U_UNASSIGNED},
{0x01f700, 0x01f776, PG_U_OTHER_SYMBOL},
- {0x01f777, 0x01f77a, PG_U_UNASSIGNED},
{0x01f77b, 0x01f7d9, PG_U_OTHER_SYMBOL},
- {0x01f7da, 0x01f7df, PG_U_UNASSIGNED},
{0x01f7e0, 0x01f7eb, PG_U_OTHER_SYMBOL},
- {0x01f7ec, 0x01f7ef, PG_U_UNASSIGNED},
{0x01f7f0, 0x01f7f0, PG_U_OTHER_SYMBOL},
- {0x01f7f1, 0x01f7ff, PG_U_UNASSIGNED},
{0x01f800, 0x01f80b, PG_U_OTHER_SYMBOL},
- {0x01f80c, 0x01f80f, PG_U_UNASSIGNED},
{0x01f810, 0x01f847, PG_U_OTHER_SYMBOL},
- {0x01f848, 0x01f84f, PG_U_UNASSIGNED},
{0x01f850, 0x01f859, PG_U_OTHER_SYMBOL},
- {0x01f85a, 0x01f85f, PG_U_UNASSIGNED},
{0x01f860, 0x01f887, PG_U_OTHER_SYMBOL},
- {0x01f888, 0x01f88f, PG_U_UNASSIGNED},
{0x01f890, 0x01f8ad, PG_U_OTHER_SYMBOL},
- {0x01f8ae, 0x01f8af, PG_U_UNASSIGNED},
{0x01f8b0, 0x01f8b1, PG_U_OTHER_SYMBOL},
- {0x01f8b2, 0x01f8ff, PG_U_UNASSIGNED},
{0x01f900, 0x01fa53, PG_U_OTHER_SYMBOL},
- {0x01fa54, 0x01fa5f, PG_U_UNASSIGNED},
{0x01fa60, 0x01fa6d, PG_U_OTHER_SYMBOL},
- {0x01fa6e, 0x01fa6f, PG_U_UNASSIGNED},
{0x01fa70, 0x01fa7c, PG_U_OTHER_SYMBOL},
- {0x01fa7d, 0x01fa7f, PG_U_UNASSIGNED},
{0x01fa80, 0x01fa88, PG_U_OTHER_SYMBOL},
- {0x01fa89, 0x01fa8f, PG_U_UNASSIGNED},
{0x01fa90, 0x01fabd, PG_U_OTHER_SYMBOL},
- {0x01fabe, 0x01fabe, PG_U_UNASSIGNED},
{0x01fabf, 0x01fac5, PG_U_OTHER_SYMBOL},
- {0x01fac6, 0x01facd, PG_U_UNASSIGNED},
{0x01face, 0x01fadb, PG_U_OTHER_SYMBOL},
- {0x01fadc, 0x01fadf, PG_U_UNASSIGNED},
{0x01fae0, 0x01fae8, PG_U_OTHER_SYMBOL},
- {0x01fae9, 0x01faef, PG_U_UNASSIGNED},
{0x01faf0, 0x01faf8, PG_U_OTHER_SYMBOL},
- {0x01faf9, 0x01faff, PG_U_UNASSIGNED},
{0x01fb00, 0x01fb92, PG_U_OTHER_SYMBOL},
- {0x01fb93, 0x01fb93, PG_U_UNASSIGNED},
{0x01fb94, 0x01fbca, PG_U_OTHER_SYMBOL},
- {0x01fbcb, 0x01fbef, PG_U_UNASSIGNED},
{0x01fbf0, 0x01fbf9, PG_U_DECIMAL_NUMBER},
- {0x01fbfa, 0x01ffff, PG_U_UNASSIGNED},
{0x020000, 0x02a6df, PG_U_OTHER_LETTER},
- {0x02a6e0, 0x02a6ff, PG_U_UNASSIGNED},
{0x02a700, 0x02b739, PG_U_OTHER_LETTER},
- {0x02b73a, 0x02b73f, PG_U_UNASSIGNED},
{0x02b740, 0x02b81d, PG_U_OTHER_LETTER},
- {0x02b81e, 0x02b81f, PG_U_UNASSIGNED},
{0x02b820, 0x02cea1, PG_U_OTHER_LETTER},
- {0x02cea2, 0x02ceaf, PG_U_UNASSIGNED},
{0x02ceb0, 0x02ebe0, PG_U_OTHER_LETTER},
- {0x02ebe1, 0x02ebef, PG_U_UNASSIGNED},
{0x02ebf0, 0x02ee5d, PG_U_OTHER_LETTER},
- {0x02ee5e, 0x02f7ff, PG_U_UNASSIGNED},
{0x02f800, 0x02fa1d, PG_U_OTHER_LETTER},
- {0x02fa1e, 0x02ffff, PG_U_UNASSIGNED},
{0x030000, 0x03134a, PG_U_OTHER_LETTER},
- {0x03134b, 0x03134f, PG_U_UNASSIGNED},
{0x031350, 0x0323af, PG_U_OTHER_LETTER},
- {0x0323b0, 0x0e0000, PG_U_UNASSIGNED},
{0x0e0001, 0x0e0001, PG_U_FORMAT},
- {0x0e0002, 0x0e001f, PG_U_UNASSIGNED},
{0x0e0020, 0x0e007f, PG_U_FORMAT},
- {0x0e0080, 0x0e00ff, PG_U_UNASSIGNED},
{0x0e0100, 0x0e01ef, PG_U_NONSPACING_MARK},
- {0x0e01f0, 0x0effff, PG_U_UNASSIGNED},
{0x0f0000, 0x0ffffd, PG_U_PRIVATE_USE},
- {0x0ffffe, 0x0fffff, PG_U_UNASSIGNED},
- {0x100000, 0x10fffd, PG_U_PRIVATE_USE},
- {0x10fffe, 0x10ffff, PG_U_UNASSIGNED}
+ {0x100000, 0x10fffd, PG_U_PRIVATE_USE}
};
--
2.34.1
[text/x-patch] v2-0001-Minor-cleanup-for-unicode-update-build-and-test.patch (7.4K, ../../[email protected]/6-v2-0001-Minor-cleanup-for-unicode-update-build-and-test.patch)
download | inline diff:
From 80ed701721b2bc91f2346f013d58930cd1d325f5 Mon Sep 17 00:00:00 2001
From: Jeff Davis <[email protected]>
Date: Wed, 22 Nov 2023 10:38:46 -0800
Subject: [PATCH v2 1/3] Minor cleanup for unicode-update build and test.
---
src/common/unicode/Makefile | 6 ++--
src/common/unicode/category_test.c | 18 ++++++------
src/common/unicode/meson.build | 44 +++++++++++++++---------------
3 files changed, 34 insertions(+), 34 deletions(-)
diff --git a/src/common/unicode/Makefile b/src/common/unicode/Makefile
index 30cd75cc6a..04d81dd5cb 100644
--- a/src/common/unicode/Makefile
+++ b/src/common/unicode/Makefile
@@ -21,7 +21,7 @@ CPPFLAGS += $(ICU_CFLAGS)
# By default, do nothing.
all:
-update-unicode: unicode_category_table.h unicode_norm_table.h unicode_nonspacing_table.h unicode_east_asian_fw_table.h unicode_normprops_table.h unicode_norm_hashfunc.h unicode_version.h
+update-unicode: unicode_category_table.h unicode_east_asian_fw_table.h unicode_nonspacing_table.h unicode_norm_hashfunc.h unicode_norm_table.h unicode_normprops_table.h unicode_version.h
mv $^ $(top_srcdir)/src/include/common/
$(MAKE) category-check
$(MAKE) normalization-check
@@ -29,7 +29,7 @@ update-unicode: unicode_category_table.h unicode_norm_table.h unicode_nonspacing
# These files are part of the Unicode Character Database. Download
# them on demand. The dependency on Makefile.global is for
# UNICODE_VERSION.
-UnicodeData.txt EastAsianWidth.txt DerivedNormalizationProps.txt CompositionExclusions.txt NormalizationTest.txt: $(top_builddir)/src/Makefile.global
+CompositionExclusions.txt DerivedNormalizationProps.txt EastAsianWidth.txt NormalizationTest.txt UnicodeData.txt: $(top_builddir)/src/Makefile.global
$(DOWNLOAD) https://www.unicode.org/Public/$(UNICODE_VERSION)/ucd/$(@F)
unicode_version.h: generate-unicode_version.pl
@@ -82,4 +82,4 @@ clean:
rm -f $(OBJS) category_test category_test.o norm_test norm_test.o
distclean: clean
- rm -f UnicodeData.txt EastAsianWidth.txt CompositionExclusions.txt NormalizationTest.txt norm_test_table.h unicode_norm_table.h
+ rm -f CompositionExclusions.txt DerivedNormalizationProps.txt EastAsianWidth.txt NormalizationTest.txt UnicodeData.txt norm_test_table.h unicode_category_table.h unicode_norm_table.h
diff --git a/src/common/unicode/category_test.c b/src/common/unicode/category_test.c
index ba62716d45..d9ea806eb8 100644
--- a/src/common/unicode/category_test.c
+++ b/src/common/unicode/category_test.c
@@ -54,8 +54,8 @@ main(int argc, char **argv)
int pg_skipped_codepoints = 0;
int icu_skipped_codepoints = 0;
- printf("Postgres Unicode Version:\t%s\n", PG_UNICODE_VERSION);
- printf("ICU Unicode Version:\t\t%s\n", U_UNICODE_VERSION);
+ printf("category_test: Postgres Unicode version:\t%s\n", PG_UNICODE_VERSION);
+ printf("category_test: ICU Unicode version:\t\t%s\n", U_UNICODE_VERSION);
for (UChar32 code = 0; code <= 0x10ffff; code++)
{
@@ -79,11 +79,11 @@ main(int argc, char **argv)
icu_skipped_codepoints++;
else
{
- printf("FAILURE for codepoint %06x\n", code);
- printf("Postgres category: %02d %s %s\n", pg_category,
+ printf("category_test: FAILURE for codepoint 0x%06x\n", code);
+ printf("category_test: Postgres category: %02d %s %s\n", pg_category,
unicode_category_abbrev(pg_category),
unicode_category_string(pg_category));
- printf("ICU category: %02d %s %s\n", icu_category,
+ printf("category_test: ICU category: %02d %s %s\n", icu_category,
unicode_category_abbrev(icu_category),
unicode_category_string(icu_category));
printf("\n");
@@ -93,16 +93,16 @@ main(int argc, char **argv)
}
if (pg_skipped_codepoints > 0)
- printf("Skipped %d codepoints unassigned in Postgres due to Unicode version mismatch.\n",
+ printf("category_test: skipped %d codepoints unassigned in Postgres due to Unicode version mismatch\n",
pg_skipped_codepoints);
if (icu_skipped_codepoints > 0)
- printf("Skipped %d codepoints unassigned in ICU due to Unicode version mismatch.\n",
+ printf("category_test: skipped %d codepoints unassigned in ICU due to Unicode version mismatch\n",
icu_skipped_codepoints);
- printf("category_test: All tests successful!\n");
+ printf("category_test: success\n");
exit(0);
#else
- printf("ICU support required for test; skipping.\n");
+ printf("category_test: ICU support required for test; skipping\n");
exit(0);
#endif
}
diff --git a/src/common/unicode/meson.build b/src/common/unicode/meson.build
index 6af46122c4..e8cfdc1df4 100644
--- a/src/common/unicode/meson.build
+++ b/src/common/unicode/meson.build
@@ -11,7 +11,7 @@ endif
# These files are part of the Unicode Character Database. Download them on
# demand.
-foreach f : ['UnicodeData.txt', 'EastAsianWidth.txt', 'DerivedNormalizationProps.txt', 'CompositionExclusions.txt', 'NormalizationTest.txt']
+foreach f : ['CompositionExclusions.txt', 'DerivedNormalizationProps.txt', 'EastAsianWidth.txt', 'NormalizationTest.txt', 'UnicodeData.txt']
url = unicode_baseurl.format(UNICODE_VERSION, f)
target = custom_target(f,
output: f,
@@ -24,15 +24,6 @@ endforeach
update_unicode_targets = []
-update_unicode_targets += \
- custom_target('unicode_version.h',
- output: ['unicode_version.h'],
- command: [
- perl, files('generate-unicode_version.pl'),
- '--outdir', '@OUTDIR@', '--version', UNICODE_VERSION],
- build_by_default: false,
- )
-
update_unicode_targets += \
custom_target('unicode_category_table.h',
input: [unicode_data['UnicodeData.txt']],
@@ -44,14 +35,12 @@ update_unicode_targets += \
)
update_unicode_targets += \
- custom_target('unicode_norm_table.h',
- input: [unicode_data['UnicodeData.txt'], unicode_data['CompositionExclusions.txt']],
- output: ['unicode_norm_table.h', 'unicode_norm_hashfunc.h'],
- depend_files: perfect_hash_pm,
- command: [
- perl, files('generate-unicode_norm_table.pl'),
- '--outdir', '@OUTDIR@', '@INPUT@'],
+ custom_target('unicode_east_asian_fw_table.h',
+ input: [unicode_data['EastAsianWidth.txt']],
+ output: ['unicode_east_asian_fw_table.h'],
+ command: [perl, files('generate-unicode_east_asian_fw_table.pl'), '@INPUT@'],
build_by_default: false,
+ capture: true,
)
update_unicode_targets += \
@@ -65,12 +54,14 @@ update_unicode_targets += \
)
update_unicode_targets += \
- custom_target('unicode_east_asian_fw_table.h',
- input: [unicode_data['EastAsianWidth.txt']],
- output: ['unicode_east_asian_fw_table.h'],
- command: [perl, files('generate-unicode_east_asian_fw_table.pl'), '@INPUT@'],
+ custom_target('unicode_norm_table.h',
+ input: [unicode_data['UnicodeData.txt'], unicode_data['CompositionExclusions.txt']],
+ output: ['unicode_norm_table.h', 'unicode_norm_hashfunc.h'],
+ depend_files: perfect_hash_pm,
+ command: [
+ perl, files('generate-unicode_norm_table.pl'),
+ '--outdir', '@OUTDIR@', '@INPUT@'],
build_by_default: false,
- capture: true,
)
update_unicode_targets += \
@@ -83,6 +74,15 @@ update_unicode_targets += \
capture: true,
)
+update_unicode_targets += \
+ custom_target('unicode_version.h',
+ output: ['unicode_version.h'],
+ command: [
+ perl, files('generate-unicode_version.pl'),
+ '--outdir', '@OUTDIR@', '--version', UNICODE_VERSION],
+ build_by_default: false,
+ )
+
norm_test_table = custom_target('norm_test_table.h',
input: [unicode_data['NormalizationTest.txt']],
output: ['norm_test_table.h'],
--
2.34.1
^ permalink raw reply [nested|flat] 57+ messages in thread
* Re: encoding affects ICU regex character classification
2023-11-30 00:23 Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-11-30 02:10 ` Re: encoding affects ICU regex character classification Thomas Munro <[email protected]>
2023-12-01 20:49 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
@ 2023-12-09 21:39 ` Thomas Munro <[email protected]>
2023-12-12 21:39 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
0 siblings, 1 reply; 57+ messages in thread
From: Thomas Munro @ 2023-12-09 21:39 UTC (permalink / raw)
To: Jeff Davis <[email protected]>; +Cc: Tom Lane <[email protected]>; pgsql-hackers
On Sat, Dec 2, 2023 at 9:49 AM Jeff Davis <[email protected]> wrote:
> Your definition is too wide in my opinion, because it mixes together
> different sources of variation that are best left separate:
> a. region/language
> b. technical requirements
> c. versioning
> d. implementation variance
>
> (a) is not a true source of variation (please correct me if I'm wrong)
>
> (b) is perhaps interesting. The "C" locale is one example, and perhaps
> there are others, but I doubt very many others that we want to support.
>
> (c) is not a major concern in my opinion. The impact of Unicode changes
> is usually not dramatic, and it only affects regexes so it's much more
> contained than collation, for example. And if you really care, just use
> the "C" locale.
>
> (d) is mostly a bug
I get you. I was mainly commenting on what POSIX APIs allow, which is
much wider than what you might observe on <your local libc>, and also
end-user-customisable. But I agree that Unicode is all-pervasive and
authoritative in practice, to the point that if your libc disagrees
with it, it's probably just wrong. (I guess site-local locales were
essential for bootstrapping in the early days of computers in a
language/territory but I can't find much discussion of the tools being
used by non-libc-maintainers today.)
> I think we only need 2 main character classification schemes: "C" and
> Unicode (TR #18 Compatibility Properties[1], either the "Standard"
> variant or the "POSIX Compatible" variant or both). The libc and ICU
> ones should be there only for compatibility and discouraged and
> hopefully eventually removed.
How would you specify what you want? As with collating, I like the
idea of keeping support for libc even if it is terrible (some libcs
more than others) and eventually not the default, because I think
optional agreement with other software on the same host is a feature.
In the regex code we see not only class membership tests eg
iswlower_l(), but also conversions eg towlower_l(). Unless you also
implement built-in case mapping, you'd still have to call libc or ICU
for that, right? It seems a bit strange to use different systems for
classification and mapping. If you do implement mapping too, you have
to decide if you believe it is language-dependent or not, I think?
Hmm, let's see what we're doing now... for ICU the regex code is using
"simple" case mapping functions like u_toupper(c) that don't take a
locale, so no Turkish i/İ conversion for you, unlike our SQL
upper()/lower(), which this is supposed to agree with according to the
comments at the top. I see why: POSIX can only do one-by-one
character mappings (which cannot handle Greek's context-sensitive
Σ->σ/ς or German's multi-character ß->SS), while ICU offers only
language-aware "full" string conversation (which does not guarantee
1:1 mapping for each character in a string) OR non-language-aware
"simple" character conversion (which does not handle Turkish's i->İ).
ICU has no middle ground for language-aware mapping with just the 1:1
cases only, probably because that doesn't really make total sense as a
concept (as I assume Greek speakers would agree).
> > > Not knowing anything about how glibc generates its charmaps,
> > > Unicode
> > > or pre-Unicode, I could take a wild guess that maybe in LATIN9 they
> > > have an old hand-crafted table, but for UTF-8 encoding it's fully
> > > outsourced to Unicode, and that's why you see a difference.
>
> No, the problem is that we're passing a pg_wchar to an ICU function
> that expects a 32-bit code point. Those two things are equivalent in
> the UTF8 encoding, but not in the LATIN9 encoding.
Ah right, I get that now (sorry, I confused myself by forgetting we
were talking about ICU).
^ permalink raw reply [nested|flat] 57+ messages in thread
* Re: encoding affects ICU regex character classification
2023-11-30 00:23 Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-11-30 02:10 ` Re: encoding affects ICU regex character classification Thomas Munro <[email protected]>
2023-12-01 20:49 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-12-09 21:39 ` Re: encoding affects ICU regex character classification Thomas Munro <[email protected]>
@ 2023-12-12 21:39 ` Jeff Davis <[email protected]>
2023-12-12 22:35 ` Re: encoding affects ICU regex character classification Jeremy Schneider <[email protected]>
0 siblings, 1 reply; 57+ messages in thread
From: Jeff Davis @ 2023-12-12 21:39 UTC (permalink / raw)
To: Thomas Munro <[email protected]>; +Cc: Tom Lane <[email protected]>; pgsql-hackers
On Sun, 2023-12-10 at 10:39 +1300, Thomas Munro wrote:
>
> How would you specify what you want?
One proposal would be to have a builtin collation provider:
https://postgr.es/m/[email protected]
I don't think there are very many ctype options, but they could be
specified as part of the locale, or perhaps even as some provider-
specific options specified at CREATE COLLATION time.
> As with collating, I like the
> idea of keeping support for libc even if it is terrible (some libcs
> more than others) and eventually not the default, because I think
> optional agreement with other software on the same host is a feature.
Of course we should keep the libc support around. I'm not sure how
relevant such a feature is, but I don't think we actually have to
remove it.
> Unless you also
> implement built-in case mapping, you'd still have to call libc or ICU
> for that, right?
We can do built-in case mapping, see:
https://postgr.es/m/[email protected]
> It seems a bit strange to use different systems for
> classification and mapping. If you do implement mapping too, you
> have
> to decide if you believe it is language-dependent or not, I think?
A complete solution would need to do the language-dependent case
mapping. But that seems to only be 3 locales ("az", "lt", and "tr"),
and only a handful of mapping changes, so we can handle that with the
builtin provider as well.
> Hmm, let's see what we're doing now... for ICU the regex code is
> using
> "simple" case mapping functions like u_toupper(c) that don't take a
> locale, so no Turkish i/İ conversion for you, unlike our SQL
> upper()/lower(), which this is supposed to agree with according to
> the
> comments at the top. I see why: POSIX can only do one-by-one
> character mappings (which cannot handle Greek's context-sensitive
> Σ->σ/ς or German's multi-character ß->SS)
Regexes are inherently character-by-character, so transformations like
ß->SS are not going to work for case-insensitive regex matching
regardless of the provider.
Σ->σ/ς does make sense, and what we have seems to be just broken:
select 'ς' ~* 'Σ'; -- false in both libc and ICU
select 'Σ' ~* 'ς'; -- true in both libc and ICU
Similarly for titlecase variants:
select 'Dž' ~* 'dž'; -- false in libc and ICU
select 'dž' ~* 'Dž'; -- true in libc and ICU
If we do the case mapping ourselves, we can make those work. We'd just
have to modify the APIs a bit so that allcases() can actually get all
of the case variants, rather than relying on just towupper/towlower.
Regards,
Jeff Davis
^ permalink raw reply [nested|flat] 57+ messages in thread
* Re: encoding affects ICU regex character classification
2023-11-30 00:23 Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-11-30 02:10 ` Re: encoding affects ICU regex character classification Thomas Munro <[email protected]>
2023-12-01 20:49 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-12-09 21:39 ` Re: encoding affects ICU regex character classification Thomas Munro <[email protected]>
2023-12-12 21:39 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
@ 2023-12-12 22:35 ` Jeremy Schneider <[email protected]>
2023-12-14 15:12 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
0 siblings, 1 reply; 57+ messages in thread
From: Jeremy Schneider @ 2023-12-12 22:35 UTC (permalink / raw)
To: Jeff Davis <[email protected]>; Thomas Munro <[email protected]>; +Cc: Tom Lane <[email protected]>; pgsql-hackers
On 12/12/23 1:39 PM, Jeff Davis wrote:
> On Sun, 2023-12-10 at 10:39 +1300, Thomas Munro wrote:
>> Unless you also
>> implement built-in case mapping, you'd still have to call libc or ICU
>> for that, right?
>
> We can do built-in case mapping, see:
>
> https://postgr.es/m/[email protected]
>
>> It seems a bit strange to use different systems for
>> classification and mapping. If you do implement mapping too, you
>> have
>> to decide if you believe it is language-dependent or not, I think?
>
> A complete solution would need to do the language-dependent case
> mapping. But that seems to only be 3 locales ("az", "lt", and "tr"),
> and only a handful of mapping changes, so we can handle that with the
> builtin provider as well.
This thread has me second-guessing the reply I just sent on the other
thread.
Is someone able to test out upper & lower functions on U+A7BA ... U+A7BF
across a few libs/versions? Theoretically the upper/lower behavior
should change in ICU between Ubuntu 18.04 LTS and Ubuntu 20.04 LTS
(specifically in ICU 64 / Unicode 12). And I have no idea if or when
glibc might have picked up the new unicode characters.
-Jeremy
--
http://about.me/jeremy_schneider
^ permalink raw reply [nested|flat] 57+ messages in thread
* Re: encoding affects ICU regex character classification
2023-11-30 00:23 Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-11-30 02:10 ` Re: encoding affects ICU regex character classification Thomas Munro <[email protected]>
2023-12-01 20:49 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-12-09 21:39 ` Re: encoding affects ICU regex character classification Thomas Munro <[email protected]>
2023-12-12 21:39 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-12-12 22:35 ` Re: encoding affects ICU regex character classification Jeremy Schneider <[email protected]>
@ 2023-12-14 15:12 ` Jeff Davis <[email protected]>
2023-12-16 00:48 ` Re: encoding affects ICU regex character classification Jeremy Schneider <[email protected]>
0 siblings, 1 reply; 57+ messages in thread
From: Jeff Davis @ 2023-12-14 15:12 UTC (permalink / raw)
To: Jeremy Schneider <[email protected]>; Thomas Munro <[email protected]>; +Cc: Tom Lane <[email protected]>; pgsql-hackers
On Tue, 2023-12-12 at 14:35 -0800, Jeremy Schneider wrote:
> Is someone able to test out upper & lower functions on U+A7BA ...
> U+A7BF
> across a few libs/versions?
Those code points are unassigned in Unicode 11.0 and assigned in
Unicode 12.0.
In ICU 63-2 (based on Unicode 11.0), they just get mapped to
themselves. In ICU 64-2 (based on Unicode 12.1) they get mapped the
same way the builtin CTYPE maps them (based on Unicode 15.1).
The concern over unassigned code points is misplaced. The application
may be aware of newly-assigned code points, and there's no way they
will be mapped correctly in Postgres if the provider is not aware of
those code points. The user can either proceed in using unassigned code
points and accept the risk of future changes, or wait for the provider
to be upgraded.
If the user doesn't have many expression indexes dependent on ctype
behavior, it doesn't matter much. If they do have such indexes, the
best we can offer is a controlled process, and the builtin provider
allows the most visibility and control.
(Aside: case mapping has very strong compatibility guarantees, but not
perfect. For better compatibility guarantees, we should support case
folding.)
> And I have no idea if or when
> glibc might have picked up the new unicode characters.
That's a strong argument in favor of a builtin provider.
Regards,
Jeff Davis
^ permalink raw reply [nested|flat] 57+ messages in thread
* Re: encoding affects ICU regex character classification
2023-11-30 00:23 Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-11-30 02:10 ` Re: encoding affects ICU regex character classification Thomas Munro <[email protected]>
2023-12-01 20:49 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-12-09 21:39 ` Re: encoding affects ICU regex character classification Thomas Munro <[email protected]>
2023-12-12 21:39 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-12-12 22:35 ` Re: encoding affects ICU regex character classification Jeremy Schneider <[email protected]>
2023-12-14 15:12 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
@ 2023-12-16 00:48 ` Jeremy Schneider <[email protected]>
2023-12-18 20:39 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
0 siblings, 1 reply; 57+ messages in thread
From: Jeremy Schneider @ 2023-12-16 00:48 UTC (permalink / raw)
To: Jeff Davis <[email protected]>; Thomas Munro <[email protected]>; +Cc: Tom Lane <[email protected]>; pgsql-hackers
On 12/14/23 7:12 AM, Jeff Davis wrote:
> The concern over unassigned code points is misplaced. The application
> may be aware of newly-assigned code points, and there's no way they
> will be mapped correctly in Postgres if the provider is not aware of
> those code points. The user can either proceed in using unassigned code
> points and accept the risk of future changes, or wait for the provider
> to be upgraded.
This does not seem to me like a good way to view the situation.
Earlier this summer, a day or two after writing a document, I was
completely surprised to open it on my work computer and see "unknown
character" boxes. When I had previously written the document on my home
computer and when I had viewed it from my cell phone, everything was
fine. Apple does a very good job of always keeping iPhones and MacOS
versions up-to-date with the latest versions of Unicode and latest
characters. iPhone keyboards make it very easy to access any character.
Emojis are the canonical example here. My work computer was one major
version of MacOS behind my home computer.
And I'm probably one of a few people on this hackers email list who even
understands what the words "unassigned code point" mean. Generally DBAs,
sysadmins, architects and developers who are all part of the tangled web
of building and maintaining systems which use PostgreSQL on their
backend are never going to think about unicode characters proactively.
This goes back to my other thread (which sadly got very little
discussion): PosgreSQL really needs to be safe by /default/ ... having
GUCs is fine though; we can put explanation in the docs about what users
should consider if they change a setting.
-Jeremy
--
http://about.me/jeremy_schneider
^ permalink raw reply [nested|flat] 57+ messages in thread
* Re: encoding affects ICU regex character classification
2023-11-30 00:23 Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-11-30 02:10 ` Re: encoding affects ICU regex character classification Thomas Munro <[email protected]>
2023-12-01 20:49 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-12-09 21:39 ` Re: encoding affects ICU regex character classification Thomas Munro <[email protected]>
2023-12-12 21:39 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-12-12 22:35 ` Re: encoding affects ICU regex character classification Jeremy Schneider <[email protected]>
2023-12-14 15:12 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-12-16 00:48 ` Re: encoding affects ICU regex character classification Jeremy Schneider <[email protected]>
@ 2023-12-18 20:39 ` Jeff Davis <[email protected]>
0 siblings, 0 replies; 57+ messages in thread
From: Jeff Davis @ 2023-12-18 20:39 UTC (permalink / raw)
To: Jeremy Schneider <[email protected]>; Thomas Munro <[email protected]>; +Cc: Tom Lane <[email protected]>; pgsql-hackers
On Fri, 2023-12-15 at 16:48 -0800, Jeremy Schneider wrote:
> This goes back to my other thread (which sadly got very little
> discussion): PosgreSQL really needs to be safe by /default/
Doesn't a built-in provider help create a safer option?
The built-in provider's version of Unicode will be consistent with
unicode_assigned(), which is a first step toward rejecting code points
that the provider doesn't understand. And by rejecting unassigned code
points, we get all kinds of Unicode compatibility guarantees that avoid
the kinds of change risks that you are worried about.
Regards,
Jeff Davis
^ permalink raw reply [nested|flat] 57+ messages in thread
end of thread, other threads:[~2023-12-18 20:39 UTC | newest]
Thread overview: 57+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v3 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v6 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v8 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v8 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v4] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v5 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-08-03 23:21 [PATCH v1 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2020-09-10 20:50 [PATCH v2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2021-03-25 15:44 [PATCH v9 2/2] ALTER TABLE ... DETACH CONCURRENTLY Alvaro Herrera <[email protected]>
2023-11-30 00:23 Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-11-30 02:10 ` Re: encoding affects ICU regex character classification Thomas Munro <[email protected]>
2023-12-01 20:49 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-12-09 21:39 ` Re: encoding affects ICU regex character classification Thomas Munro <[email protected]>
2023-12-12 21:39 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-12-12 22:35 ` Re: encoding affects ICU regex character classification Jeremy Schneider <[email protected]>
2023-12-14 15:12 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
2023-12-16 00:48 ` Re: encoding affects ICU regex character classification Jeremy Schneider <[email protected]>
2023-12-18 20:39 ` Re: encoding affects ICU regex character classification Jeff Davis <[email protected]>
This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox