public inbox for [email protected]
help / color / mirror / Atom feedFrom: Maxim Orlov <[email protected]>
To: Heikki Linnakangas <[email protected]>
Cc: wenhui qiu <[email protected]>
Cc: Alexander Korotkov <[email protected]>
Cc: Postgres hackers <[email protected]>
Subject: Re: POC: make mxidoff 64 bits
Date: Mon, 18 Nov 2024 16:22:38 +0300
Message-ID: <CACG=ezZGQFBb0yepka8hU2BmJ48ujt3xa+aYLNL0BQPx0vqwZg@mail.gmail.com> (raw)
In-Reply-To: <CACG=ezb680eb=JXh1ns=t5eGH3h9y-uTfT4tf3Xc8t2UH2q6tQ@mail.gmail.com>
References: <CACG=ezaWg7_nt-8ey4aKv2w9LcuLthHknwCawmBgEeTnJrJTcw@mail.gmail.com>
<CACG=ezYokoiumOFnqUfg_ffHD5s8T+6iHYfzKLfa=QQ-1pNrBg@mail.gmail.com>
<CACG=ezY9xq73jcX_EjVqx5-f90nbQ9PyhFCTW2fwFCS2wmNiFw@mail.gmail.com>
<CACG=eza+27CfLBobJJccRhXrA3He6c1irAnoyTtSC1-z9UXLrg@mail.gmail.com>
<CAPpHfduczcop9s6gKUpLGgFUe2y4ERGMJx6SS6Kp+s-kQPwMjg@mail.gmail.com>
<CACG=ezbye4g_ERNqE=gBcvQ0YypRaVENhNUu8xrs4PL12UdnUA@mail.gmail.com>
<CACG=ezaMncd0-BcGHBgsSR2eqHfrz9WznHGLKX8biz6zu-azGw@mail.gmail.com>
<[email protected]>
<CACG=ezb9XTvd3ZmS0y8gUunx_wBBdJO7ou+BfCOnnA5jE-11vg@mail.gmail.com>
<CACG=ezYFNqGjsxF6Vb2CHF6JzKcjhAFauaFm9js0nu_3Ngcdkw@mail.gmail.com>
<CAGjGUA+dcV7veaCV1H65vCNsbS++nT8=ho772gDvsXUW9H7eXQ@mail.gmail.com>
<CACG=ezYThNkf8QsDA-aQfEFEkqn2L=_uUL83z0vJstPRasbZqg@mail.gmail.com>
<[email protected]>
<CACG=ezYtCatcRODS-ZkwhcxuqBKCuhEsZGBruw=dGCLoepF+ZA@mail.gmail.com>
<[email protected]>
<CACG=ezb680eb=JXh1ns=t5eGH3h9y-uTfT4tf3Xc8t2UH2q6tQ@mail.gmail.com>
Shame on me! I've sent an erroneous patch set. Version 7 is defective. Here
is the proper version v8 with minor refactoring in segresize.c.
Also, I rename bump cat version patch into txt in order not to break cfbot.
--
Best regards,
Maxim Orlov.
From 73b8663093ff1c58def9a80abab142a12c993bf6 Mon Sep 17 00:00:00 2001
From: Maxim Orlov <[email protected]>
Date: Wed, 13 Nov 2024 16:34:34 +0300
Subject: [PATCH v8 5/5] TEST: bump catver
---
src/bin/pg_upgrade/pg_upgrade.h | 2 +-
src/include/catalog/catversion.h | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 2c85ec1e94..18faedc963 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -119,7 +119,7 @@ extern char *output_files[];
*
* XXX: should be changed to the actual CATALOG_VERSION_NO on commit.
*/
-#define MULTIXACTOFFSET_FORMATCHANGE_CAT_VER 202409041
+#define MULTIXACTOFFSET_FORMATCHANGE_CAT_VER 202411112
/*
* large object chunk size added to pg_controldata,
diff --git a/src/include/catalog/catversion.h b/src/include/catalog/catversion.h
index 5dd91e190a..3d09caf5ae 100644
--- a/src/include/catalog/catversion.h
+++ b/src/include/catalog/catversion.h
@@ -57,6 +57,6 @@
*/
/* yyyymmddN */
-#define CATALOG_VERSION_NO 202411111
+#define CATALOG_VERSION_NO 202411112
#endif
--
2.43.0
Attachments:
[text/plain] v8-0005-TEST-bump-catver.patch.txt (1.1K, 3-v8-0005-TEST-bump-catver.patch.txt)
download | inline diff:
From 73b8663093ff1c58def9a80abab142a12c993bf6 Mon Sep 17 00:00:00 2001
From: Maxim Orlov <[email protected]>
Date: Wed, 13 Nov 2024 16:34:34 +0300
Subject: [PATCH v8 5/5] TEST: bump catver
---
src/bin/pg_upgrade/pg_upgrade.h | 2 +-
src/include/catalog/catversion.h | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 2c85ec1e94..18faedc963 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -119,7 +119,7 @@ extern char *output_files[];
*
* XXX: should be changed to the actual CATALOG_VERSION_NO on commit.
*/
-#define MULTIXACTOFFSET_FORMATCHANGE_CAT_VER 202409041
+#define MULTIXACTOFFSET_FORMATCHANGE_CAT_VER 202411112
/*
* large object chunk size added to pg_controldata,
diff --git a/src/include/catalog/catversion.h b/src/include/catalog/catversion.h
index 5dd91e190a..3d09caf5ae 100644
--- a/src/include/catalog/catversion.h
+++ b/src/include/catalog/catversion.h
@@ -57,6 +57,6 @@
*/
/* yyyymmddN */
-#define CATALOG_VERSION_NO 202411111
+#define CATALOG_VERSION_NO 202411112
#endif
--
2.43.0
[application/octet-stream] v8-0002-Use-64-bit-multixact-offsets.patch (13.3K, 4-v8-0002-Use-64-bit-multixact-offsets.patch)
download | inline diff:
From ad9a1509fd5cd68838169b3465ab4c5f9827a4e3 Mon Sep 17 00:00:00 2001
From: Maxim Orlov <[email protected]>
Date: Wed, 6 Mar 2024 11:11:33 +0300
Subject: [PATCH v8 2/5] Use 64-bit multixact offsets.
Author: Maxim Orlov <[email protected]>
---
src/backend/access/transam/multixact.c | 172 +------------------------
src/bin/pg_resetwal/pg_resetwal.c | 2 +-
src/bin/pg_resetwal/t/001_basic.pl | 2 +-
src/include/access/multixact.h | 2 +-
src/include/c.h | 2 +-
5 files changed, 11 insertions(+), 169 deletions(-)
diff --git a/src/backend/access/transam/multixact.c b/src/backend/access/transam/multixact.c
index ab90912ed3..c51e03e832 100644
--- a/src/backend/access/transam/multixact.c
+++ b/src/backend/access/transam/multixact.c
@@ -96,14 +96,6 @@
/*
* Defines for MultiXactOffset page sizes. A page is the same BLCKSZ as is
* used everywhere else in Postgres.
- *
- * Note: because MultiXactOffsets are 32 bits and wrap around at 0xFFFFFFFF,
- * MultiXact page numbering also wraps around at
- * 0xFFFFFFFF/MULTIXACT_OFFSETS_PER_PAGE, and segment numbering at
- * 0xFFFFFFFF/MULTIXACT_OFFSETS_PER_PAGE/SLRU_PAGES_PER_SEGMENT. We need
- * take no explicit notice of that fact in this module, except when comparing
- * segment and page numbers in TruncateMultiXact (see
- * MultiXactOffsetPagePrecedes).
*/
/* We need four bytes per offset */
@@ -272,9 +264,6 @@ typedef struct MultiXactStateData
MultiXactId multiStopLimit;
MultiXactId multiWrapLimit;
- /* support for members anti-wraparound measures */
- MultiXactOffset offsetStopLimit; /* known if oldestOffsetKnown */
-
/*
* This is used to sleep until a multixact offset is written when we want
* to create the next one.
@@ -409,8 +398,6 @@ static bool MultiXactOffsetPrecedes(MultiXactOffset offset1,
MultiXactOffset offset2);
static void ExtendMultiXactOffset(MultiXactId multi);
static void ExtendMultiXactMember(MultiXactOffset offset, int nmembers);
-static bool MultiXactOffsetWouldWrap(MultiXactOffset boundary,
- MultiXactOffset start, uint32 distance);
static bool SetOffsetVacuumLimit(bool is_startup);
static bool find_multixact_start(MultiXactId multi, MultiXactOffset *result);
static void WriteMZeroPageXlogRec(int64 pageno, uint8 info);
@@ -1164,78 +1151,6 @@ GetNewMultiXactId(int nmembers, MultiXactOffset *offset)
else
*offset = nextOffset;
- /*----------
- * Protect against overrun of the members space as well, with the
- * following rules:
- *
- * If we're past offsetStopLimit, refuse to generate more multis.
- * If we're close to offsetStopLimit, emit a warning.
- *
- * Arbitrarily, we start emitting warnings when we're 20 segments or less
- * from offsetStopLimit.
- *
- * Note we haven't updated the shared state yet, so if we fail at this
- * point, the multixact ID we grabbed can still be used by the next guy.
- *
- * Note that there is no point in forcing autovacuum runs here: the
- * multixact freeze settings would have to be reduced for that to have any
- * effect.
- *----------
- */
-#define OFFSET_WARN_SEGMENTS 20
- if (MultiXactState->oldestOffsetKnown &&
- MultiXactOffsetWouldWrap(MultiXactState->offsetStopLimit, nextOffset,
- nmembers))
- {
- /* see comment in the corresponding offsets wraparound case */
- SendPostmasterSignal(PMSIGNAL_START_AUTOVAC_LAUNCHER);
-
- ereport(ERROR,
- (errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
- errmsg("multixact \"members\" limit exceeded"),
- errdetail_plural("This command would create a multixact with %u members, but the remaining space is only enough for %u member.",
- "This command would create a multixact with %u members, but the remaining space is only enough for %u members.",
- MultiXactState->offsetStopLimit - nextOffset - 1,
- nmembers,
- MultiXactState->offsetStopLimit - nextOffset - 1),
- errhint("Execute a database-wide VACUUM in database with OID %u with reduced \"vacuum_multixact_freeze_min_age\" and \"vacuum_multixact_freeze_table_age\" settings.",
- MultiXactState->oldestMultiXactDB)));
- }
-
- /*
- * Check whether we should kick autovacuum into action, to prevent members
- * wraparound. NB we use a much larger window to trigger autovacuum than
- * just the warning limit. The warning is just a measure of last resort -
- * this is in line with GetNewTransactionId's behaviour.
- */
- if (!MultiXactState->oldestOffsetKnown ||
- (MultiXactState->nextOffset - MultiXactState->oldestOffset
- > MULTIXACT_MEMBER_SAFE_THRESHOLD))
- {
- /*
- * To avoid swamping the postmaster with signals, we issue the autovac
- * request only when crossing a segment boundary. With default
- * compilation settings that's roughly after 50k members. This still
- * gives plenty of chances before we get into real trouble.
- */
- if ((MXOffsetToMemberPage(nextOffset) / SLRU_PAGES_PER_SEGMENT) !=
- (MXOffsetToMemberPage(nextOffset + nmembers) / SLRU_PAGES_PER_SEGMENT))
- SendPostmasterSignal(PMSIGNAL_START_AUTOVAC_LAUNCHER);
- }
-
- if (MultiXactState->oldestOffsetKnown &&
- MultiXactOffsetWouldWrap(MultiXactState->offsetStopLimit,
- nextOffset,
- nmembers + MULTIXACT_MEMBERS_PER_PAGE * SLRU_PAGES_PER_SEGMENT * OFFSET_WARN_SEGMENTS))
- ereport(WARNING,
- (errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
- errmsg_plural("database with OID %u must be vacuumed before %d more multixact member is used",
- "database with OID %u must be vacuumed before %d more multixact members are used",
- MultiXactState->offsetStopLimit - nextOffset + nmembers,
- MultiXactState->oldestMultiXactDB,
- MultiXactState->offsetStopLimit - nextOffset + nmembers),
- errhint("Execute a database-wide VACUUM in that database with reduced \"vacuum_multixact_freeze_min_age\" and \"vacuum_multixact_freeze_table_age\" settings.")));
-
ExtendMultiXactMember(nextOffset, nmembers);
/*
@@ -1976,7 +1891,7 @@ MultiXactShmemInit(void)
"pg_multixact/offsets", LWTRANCHE_MULTIXACTOFFSET_BUFFER,
LWTRANCHE_MULTIXACTOFFSET_SLRU,
SYNC_HANDLER_MULTIXACT_OFFSET,
- false);
+ true);
SlruPagePrecedesUnitTests(MultiXactOffsetCtl, MULTIXACT_OFFSETS_PER_PAGE);
SimpleLruInit(MultiXactMemberCtl,
"multixact_member", multixact_member_buffers, 0,
@@ -2721,8 +2636,6 @@ SetOffsetVacuumLimit(bool is_startup)
MultiXactOffset nextOffset;
bool oldestOffsetKnown = false;
bool prevOldestOffsetKnown;
- MultiXactOffset offsetStopLimit = 0;
- MultiXactOffset prevOffsetStopLimit;
/*
* NB: Have to prevent concurrent truncation, we might otherwise try to
@@ -2737,7 +2650,6 @@ SetOffsetVacuumLimit(bool is_startup)
nextOffset = MultiXactState->nextOffset;
prevOldestOffsetKnown = MultiXactState->oldestOffsetKnown;
prevOldestOffset = MultiXactState->oldestOffset;
- prevOffsetStopLimit = MultiXactState->offsetStopLimit;
Assert(MultiXactState->finishedStartup);
LWLockRelease(MultiXactGenLock);
@@ -2768,11 +2680,7 @@ SetOffsetVacuumLimit(bool is_startup)
oldestOffsetKnown =
find_multixact_start(oldestMultiXactId, &oldestOffset);
- if (oldestOffsetKnown)
- ereport(DEBUG1,
- (errmsg_internal("oldest MultiXactId member is at offset %u",
- oldestOffset)));
- else
+ if (!oldestOffsetKnown)
ereport(LOG,
(errmsg("MultiXact member wraparound protections are disabled because oldest checkpointed MultiXact %u does not exist on disk",
oldestMultiXactId)));
@@ -2785,24 +2693,7 @@ SetOffsetVacuumLimit(bool is_startup)
* overrun of old data in the members SLRU area. We can only do so if the
* oldest offset is known though.
*/
- if (oldestOffsetKnown)
- {
- /* move back to start of the corresponding segment */
- offsetStopLimit = oldestOffset - (oldestOffset %
- (MULTIXACT_MEMBERS_PER_PAGE * SLRU_PAGES_PER_SEGMENT));
-
- /* always leave one segment before the wraparound point */
- offsetStopLimit -= (MULTIXACT_MEMBERS_PER_PAGE * SLRU_PAGES_PER_SEGMENT);
-
- if (!prevOldestOffsetKnown && !is_startup)
- ereport(LOG,
- (errmsg("MultiXact member wraparound protections are now enabled")));
-
- ereport(DEBUG1,
- (errmsg_internal("MultiXact member stop limit is now %u based on MultiXact %u",
- offsetStopLimit, oldestMultiXactId)));
- }
- else if (prevOldestOffsetKnown)
+ if (prevOldestOffsetKnown)
{
/*
* If we failed to get the oldest offset this time, but we have a
@@ -2812,14 +2703,12 @@ SetOffsetVacuumLimit(bool is_startup)
*/
oldestOffset = prevOldestOffset;
oldestOffsetKnown = true;
- offsetStopLimit = prevOffsetStopLimit;
}
/* Install the computed values */
LWLockAcquire(MultiXactGenLock, LW_EXCLUSIVE);
MultiXactState->oldestOffset = oldestOffset;
MultiXactState->oldestOffsetKnown = oldestOffsetKnown;
- MultiXactState->offsetStopLimit = offsetStopLimit;
LWLockRelease(MultiXactGenLock);
/*
@@ -2829,54 +2718,6 @@ SetOffsetVacuumLimit(bool is_startup)
(nextOffset - oldestOffset > MULTIXACT_MEMBER_SAFE_THRESHOLD);
}
-/*
- * Return whether adding "distance" to "start" would move past "boundary".
- *
- * We use this to determine whether the addition is "wrapping around" the
- * boundary point, hence the name. The reason we don't want to use the regular
- * 2^31-modulo arithmetic here is that we want to be able to use the whole of
- * the 2^32-1 space here, allowing for more multixacts than would fit
- * otherwise.
- */
-static bool
-MultiXactOffsetWouldWrap(MultiXactOffset boundary, MultiXactOffset start,
- uint32 distance)
-{
- MultiXactOffset finish;
-
- /*
- * Note that offset number 0 is not used (see GetMultiXactIdMembers), so
- * if the addition wraps around the UINT_MAX boundary, skip that value.
- */
- finish = start + distance;
- if (finish < start)
- finish++;
-
- /*-----------------------------------------------------------------------
- * When the boundary is numerically greater than the starting point, any
- * value numerically between the two is not wrapped:
- *
- * <----S----B---->
- * [---) = F wrapped past B (and UINT_MAX)
- * [---) = F not wrapped
- * [----] = F wrapped past B
- *
- * When the boundary is numerically less than the starting point (i.e. the
- * UINT_MAX wraparound occurs somewhere in between) then all values in
- * between are wrapped:
- *
- * <----B----S---->
- * [---) = F not wrapped past B (but wrapped past UINT_MAX)
- * [---) = F wrapped past B (and UINT_MAX)
- * [----] = F not wrapped
- *-----------------------------------------------------------------------
- */
- if (start < boundary)
- return finish >= boundary || finish < start;
- else
- return finish >= boundary && finish < start;
-}
-
/*
* Find the starting offset of the given MultiXactId.
*
@@ -2998,8 +2839,9 @@ MultiXactMemberFreezeThreshold(void)
* we try to eliminate from the system is based on how far we are past
* MULTIXACT_MEMBER_SAFE_THRESHOLD.
*/
- fraction = (double) (members - MULTIXACT_MEMBER_SAFE_THRESHOLD) /
- (MULTIXACT_MEMBER_DANGER_THRESHOLD - MULTIXACT_MEMBER_SAFE_THRESHOLD);
+ fraction = (double) (members - MULTIXACT_MEMBER_SAFE_THRESHOLD);
+ fraction /= (double) (MULTIXACT_MEMBER_DANGER_THRESHOLD - MULTIXACT_MEMBER_SAFE_THRESHOLD);
+
victim_multixacts = multixacts * fraction;
/* fraction could be > 1.0, but lowest possible freeze age is zero */
@@ -3345,7 +3187,7 @@ MultiXactIdPrecedesOrEquals(MultiXactId multi1, MultiXactId multi2)
static bool
MultiXactOffsetPrecedes(MultiXactOffset offset1, MultiXactOffset offset2)
{
- int32 diff = (int32) (offset1 - offset2);
+ int64 diff = (int64) (offset1 - offset2);
return (diff < 0);
}
diff --git a/src/bin/pg_resetwal/pg_resetwal.c b/src/bin/pg_resetwal/pg_resetwal.c
index 985cd06802..1af2ce4b93 100644
--- a/src/bin/pg_resetwal/pg_resetwal.c
+++ b/src/bin/pg_resetwal/pg_resetwal.c
@@ -264,7 +264,7 @@ main(int argc, char *argv[])
case 'O':
errno = 0;
- set_mxoff = strtoul(optarg, &endptr, 0);
+ set_mxoff = strtou64(optarg, &endptr, 0);
if (endptr == optarg || *endptr != '\0' || errno != 0)
{
pg_log_error("invalid argument for option %s", "-O");
diff --git a/src/bin/pg_resetwal/t/001_basic.pl b/src/bin/pg_resetwal/t/001_basic.pl
index 9829e48106..f8a8eef44d 100644
--- a/src/bin/pg_resetwal/t/001_basic.pl
+++ b/src/bin/pg_resetwal/t/001_basic.pl
@@ -206,7 +206,7 @@ push @cmd,
sprintf("%d,%d", hex($files[0]) == 0 ? 3 : hex($files[0]), hex($files[-1]));
@files = get_slru_files('pg_multixact/offsets');
-$mult = 32 * $blcksz / 4;
+$mult = 32 * $blcksz / 8;
# -m argument is "new,old"
push @cmd, '-m',
sprintf("%d,%d",
diff --git a/src/include/access/multixact.h b/src/include/access/multixact.h
index 7ffd256c74..90583634ec 100644
--- a/src/include/access/multixact.h
+++ b/src/include/access/multixact.h
@@ -27,7 +27,7 @@
#define MultiXactIdIsValid(multi) ((multi) != InvalidMultiXactId)
-#define MaxMultiXactOffset ((MultiXactOffset) 0xFFFFFFFF)
+#define MaxMultiXactOffset UINT64CONST(0xFFFFFFFFFFFFFFFF)
/*
* Possible multixact lock modes ("status"). The first four modes are for
diff --git a/src/include/c.h b/src/include/c.h
index 0a548d69d7..e1b3187d0b 100644
--- a/src/include/c.h
+++ b/src/include/c.h
@@ -664,7 +664,7 @@ typedef uint32 SubTransactionId;
/* MultiXactId must be equivalent to TransactionId, to fit in t_xmax */
typedef TransactionId MultiXactId;
-typedef uint32 MultiXactOffset;
+typedef uint64 MultiXactOffset;
typedef uint32 CommandId;
--
2.43.0
[application/x-sh] generate.sh (2.1K, 5-generate.sh)
download
[application/x-sh] test.sh (2.1K, 6-test.sh)
download
[text/x-python-script] multixids.py (1.6K, 7-multixids.py)
download | inline:
#!/usr/bin/env python3
import sys;
import threading;
import psycopg2;
def test_multixact(tblname: str):
with psycopg2.connect(dbname="postgres") as conn:
cur = conn.cursor()
cur.execute(
f"""
DROP TABLE IF EXISTS {tblname};
CREATE TABLE {tblname}(i int primary key, n_updated int) WITH (autovacuum_enabled=false);
INSERT INTO {tblname} select g, 0 from generate_series(1, 50) g;
"""
)
# Lock entries using parallel connections in a round-robin fashion.
nclients = 50
update_every = 97
connections = []
for _ in range(nclients):
# Do not turn on autocommit. We want to hold the key-share locks.
conn = psycopg2.connect(dbname="postgres")
connections.append(conn)
# On each iteration, we commit the previous transaction on a connection,
# and issue another select. Each SELECT generates a new multixact that
# includes the new XID, and the XIDs of all the other parallel transactions.
# This generates enough traffic on both multixact offsets and members SLRUs
# to cross page boundaries.
for i in range(20000):
conn = connections[i % nclients]
conn.commit()
# Perform some non-key UPDATEs too, to exercise different multixact
# member statuses.
if i % update_every == 0:
conn.cursor().execute(f"update {tblname} set n_updated = n_updated + 1 where i = {i % 50}")
else:
conn.cursor().execute(f"select * from {tblname} for key share")
test_multixact(sys.argv[1])
[application/octet-stream] v8-0001-Use-64-bit-format-output-for-multixact-offsets.patch (9.0K, 8-v8-0001-Use-64-bit-format-output-for-multixact-offsets.patch)
download | inline diff:
From fdb7e2eee33dfb5df714d8d16112d4c907475d78 Mon Sep 17 00:00:00 2001
From: Maxim Orlov <[email protected]>
Date: Wed, 7 Aug 2024 16:35:22 +0300
Subject: [PATCH v8 1/5] Use 64-bit format output for multixact offsets
Author: Maxim Orlov <[email protected]>
---
src/backend/access/rmgrdesc/mxactdesc.c | 9 ++++----
src/backend/access/rmgrdesc/xlogdesc.c | 4 ++--
src/backend/access/transam/multixact.c | 26 +++++++++++++----------
src/backend/access/transam/xlogrecovery.c | 5 +++--
src/bin/pg_controldata/pg_controldata.c | 4 ++--
src/bin/pg_resetwal/pg_resetwal.c | 8 +++----
6 files changed, 31 insertions(+), 25 deletions(-)
diff --git a/src/backend/access/rmgrdesc/mxactdesc.c b/src/backend/access/rmgrdesc/mxactdesc.c
index 3e8ad4d5ef..1b486de38c 100644
--- a/src/backend/access/rmgrdesc/mxactdesc.c
+++ b/src/backend/access/rmgrdesc/mxactdesc.c
@@ -65,8 +65,8 @@ multixact_desc(StringInfo buf, XLogReaderState *record)
xl_multixact_create *xlrec = (xl_multixact_create *) rec;
int i;
- appendStringInfo(buf, "%u offset %u nmembers %d: ", xlrec->mid,
- xlrec->moff, xlrec->nmembers);
+ appendStringInfo(buf, "%u offset %llu nmembers %d: ", xlrec->mid,
+ (unsigned long long) xlrec->moff, xlrec->nmembers);
for (i = 0; i < xlrec->nmembers; i++)
out_member(buf, &xlrec->members[i]);
}
@@ -74,9 +74,10 @@ multixact_desc(StringInfo buf, XLogReaderState *record)
{
xl_multixact_truncate *xlrec = (xl_multixact_truncate *) rec;
- appendStringInfo(buf, "offsets [%u, %u), members [%u, %u)",
+ appendStringInfo(buf, "offsets [%u, %u), members [%llu, %llu)",
xlrec->startTruncOff, xlrec->endTruncOff,
- xlrec->startTruncMemb, xlrec->endTruncMemb);
+ (unsigned long long) xlrec->startTruncMemb,
+ (unsigned long long) xlrec->endTruncMemb);
}
}
diff --git a/src/backend/access/rmgrdesc/xlogdesc.c b/src/backend/access/rmgrdesc/xlogdesc.c
index 363294d623..aaa19c81c8 100644
--- a/src/backend/access/rmgrdesc/xlogdesc.c
+++ b/src/backend/access/rmgrdesc/xlogdesc.c
@@ -66,7 +66,7 @@ xlog_desc(StringInfo buf, XLogReaderState *record)
CheckPoint *checkpoint = (CheckPoint *) rec;
appendStringInfo(buf, "redo %X/%X; "
- "tli %u; prev tli %u; fpw %s; wal_level %s; xid %u:%u; oid %u; multi %u; offset %u; "
+ "tli %u; prev tli %u; fpw %s; wal_level %s; xid %u:%u; oid %u; multi %u; offset %llu; "
"oldest xid %u in DB %u; oldest multi %u in DB %u; "
"oldest/newest commit timestamp xid: %u/%u; "
"oldest running xid %u; %s",
@@ -79,7 +79,7 @@ xlog_desc(StringInfo buf, XLogReaderState *record)
XidFromFullTransactionId(checkpoint->nextXid),
checkpoint->nextOid,
checkpoint->nextMulti,
- checkpoint->nextMultiOffset,
+ (unsigned long long) checkpoint->nextMultiOffset,
checkpoint->oldestXid,
checkpoint->oldestXidDB,
checkpoint->oldestMulti,
diff --git a/src/backend/access/transam/multixact.c b/src/backend/access/transam/multixact.c
index 8c37d7eba7..ab90912ed3 100644
--- a/src/backend/access/transam/multixact.c
+++ b/src/backend/access/transam/multixact.c
@@ -1264,7 +1264,8 @@ GetNewMultiXactId(int nmembers, MultiXactOffset *offset)
LWLockRelease(MultiXactGenLock);
- debug_elog4(DEBUG2, "GetNew: returning %u offset %u", result, *offset);
+ debug_elog4(DEBUG2, "GetNew: returning %u offset %llu", result,
+ (unsigned long long) *offset);
return result;
}
@@ -2293,8 +2294,9 @@ MultiXactGetCheckptMulti(bool is_shutdown,
LWLockRelease(MultiXactGenLock);
debug_elog6(DEBUG2,
- "MultiXact: checkpoint is nextMulti %u, nextOffset %u, oldestMulti %u in DB %u",
- *nextMulti, *nextMultiOffset, *oldestMulti, *oldestMultiDB);
+ "MultiXact: checkpoint is nextMulti %u, nextOffset %llu, oldestMulti %u in DB %u",
+ *nextMulti, (unsigned long long) *nextMultiOffset, *oldestMulti,
+ *oldestMultiDB);
}
/*
@@ -2328,8 +2330,8 @@ void
MultiXactSetNextMXact(MultiXactId nextMulti,
MultiXactOffset nextMultiOffset)
{
- debug_elog4(DEBUG2, "MultiXact: setting next multi to %u offset %u",
- nextMulti, nextMultiOffset);
+ debug_elog4(DEBUG2, "MultiXact: setting next multi to %u offset %llu",
+ nextMulti, (unsigned long long) nextMultiOffset);
LWLockAcquire(MultiXactGenLock, LW_EXCLUSIVE);
MultiXactState->nextMXact = nextMulti;
MultiXactState->nextOffset = nextMultiOffset;
@@ -2519,8 +2521,8 @@ MultiXactAdvanceNextMXact(MultiXactId minMulti,
}
if (MultiXactOffsetPrecedes(MultiXactState->nextOffset, minMultiOffset))
{
- debug_elog3(DEBUG2, "MultiXact: setting next offset to %u",
- minMultiOffset);
+ debug_elog3(DEBUG2, "MultiXact: setting next offset to %llu",
+ (unsigned long long) minMultiOffset);
MultiXactState->nextOffset = minMultiOffset;
}
LWLockRelease(MultiXactGenLock);
@@ -3211,11 +3213,12 @@ TruncateMultiXact(MultiXactId newOldestMulti, Oid newOldestMultiDB)
elog(DEBUG1, "performing multixact truncation: "
"offsets [%u, %u), offsets segments [%llx, %llx), "
- "members [%u, %u), members segments [%llx, %llx)",
+ "members [%llu, %llu), members segments [%llx, %llx)",
oldestMulti, newOldestMulti,
(unsigned long long) MultiXactIdToOffsetSegment(oldestMulti),
(unsigned long long) MultiXactIdToOffsetSegment(newOldestMulti),
- oldestOffset, newOldestOffset,
+ (unsigned long long) oldestOffset,
+ (unsigned long long) newOldestOffset,
(unsigned long long) MXOffsetToMemberSegment(oldestOffset),
(unsigned long long) MXOffsetToMemberSegment(newOldestOffset));
@@ -3471,11 +3474,12 @@ multixact_redo(XLogReaderState *record)
elog(DEBUG1, "replaying multixact truncation: "
"offsets [%u, %u), offsets segments [%llx, %llx), "
- "members [%u, %u), members segments [%llx, %llx)",
+ "members [%llu, %llu), members segments [%llx, %llx)",
xlrec.startTruncOff, xlrec.endTruncOff,
(unsigned long long) MultiXactIdToOffsetSegment(xlrec.startTruncOff),
(unsigned long long) MultiXactIdToOffsetSegment(xlrec.endTruncOff),
- xlrec.startTruncMemb, xlrec.endTruncMemb,
+ (unsigned long long) xlrec.startTruncMemb,
+ (unsigned long long) xlrec.endTruncMemb,
(unsigned long long) MXOffsetToMemberSegment(xlrec.startTruncMemb),
(unsigned long long) MXOffsetToMemberSegment(xlrec.endTruncMemb));
diff --git a/src/backend/access/transam/xlogrecovery.c b/src/backend/access/transam/xlogrecovery.c
index 05c738d661..727b6e744f 100644
--- a/src/backend/access/transam/xlogrecovery.c
+++ b/src/backend/access/transam/xlogrecovery.c
@@ -876,8 +876,9 @@ InitWalRecovery(ControlFileData *ControlFile, bool *wasShutdown_ptr,
U64FromFullTransactionId(checkPoint.nextXid),
checkPoint.nextOid)));
ereport(DEBUG1,
- (errmsg_internal("next MultiXactId: %u; next MultiXactOffset: %u",
- checkPoint.nextMulti, checkPoint.nextMultiOffset)));
+ (errmsg_internal("next MultiXactId: %u; next MultiXactOffset: %llu",
+ checkPoint.nextMulti,
+ (unsigned long long) checkPoint.nextMultiOffset)));
ereport(DEBUG1,
(errmsg_internal("oldest unfrozen transaction ID: %u, in database %u",
checkPoint.oldestXid, checkPoint.oldestXidDB)));
diff --git a/src/bin/pg_controldata/pg_controldata.c b/src/bin/pg_controldata/pg_controldata.c
index 93a05d80ca..43b6727570 100644
--- a/src/bin/pg_controldata/pg_controldata.c
+++ b/src/bin/pg_controldata/pg_controldata.c
@@ -253,8 +253,8 @@ main(int argc, char *argv[])
ControlFile->checkPointCopy.nextOid);
printf(_("Latest checkpoint's NextMultiXactId: %u\n"),
ControlFile->checkPointCopy.nextMulti);
- printf(_("Latest checkpoint's NextMultiOffset: %u\n"),
- ControlFile->checkPointCopy.nextMultiOffset);
+ printf(_("Latest checkpoint's NextMultiOffset: %llu\n"),
+ (unsigned long long) ControlFile->checkPointCopy.nextMultiOffset);
printf(_("Latest checkpoint's oldestXID: %u\n"),
ControlFile->checkPointCopy.oldestXid);
printf(_("Latest checkpoint's oldestXID's DB: %u\n"),
diff --git a/src/bin/pg_resetwal/pg_resetwal.c b/src/bin/pg_resetwal/pg_resetwal.c
index e9dcb5a6d8..985cd06802 100644
--- a/src/bin/pg_resetwal/pg_resetwal.c
+++ b/src/bin/pg_resetwal/pg_resetwal.c
@@ -737,8 +737,8 @@ PrintControlValues(bool guessed)
ControlFile.checkPointCopy.nextOid);
printf(_("Latest checkpoint's NextMultiXactId: %u\n"),
ControlFile.checkPointCopy.nextMulti);
- printf(_("Latest checkpoint's NextMultiOffset: %u\n"),
- ControlFile.checkPointCopy.nextMultiOffset);
+ printf(_("Latest checkpoint's NextMultiOffset: %llu\n"),
+ (unsigned long long) ControlFile.checkPointCopy.nextMultiOffset);
printf(_("Latest checkpoint's oldestXID: %u\n"),
ControlFile.checkPointCopy.oldestXid);
printf(_("Latest checkpoint's oldestXID's DB: %u\n"),
@@ -809,8 +809,8 @@ PrintNewControlValues(void)
if (set_mxoff != -1)
{
- printf(_("NextMultiOffset: %u\n"),
- ControlFile.checkPointCopy.nextMultiOffset);
+ printf(_("NextMultiOffset: %llu\n"),
+ (unsigned long long) ControlFile.checkPointCopy.nextMultiOffset);
}
if (set_oid != 0)
--
2.43.0
[application/octet-stream] v8-0004-Get-rid-of-MultiXactMemberFreezeThreshold-call.patch (8.8K, 9-v8-0004-Get-rid-of-MultiXactMemberFreezeThreshold-call.patch)
download | inline diff:
From edff3857e1cb6c67e75be1b00fd5da1cd4bde343 Mon Sep 17 00:00:00 2001
From: Maxim Orlov <[email protected]>
Date: Wed, 23 Oct 2024 18:23:39 +0300
Subject: [PATCH v8 4/5] Get rid of MultiXactMemberFreezeThreshold call.
Since MaxMultiXactOffset are UINT64_MAX now, MULTIXACT_MEMBER_SAFE_THRESHOLD and
MULTIXACT_MEMBER_DANGER_THRESHOLD values are not meaningful any more. Thus,
MultiXactMemberFreezeThreshold is not needed too.
Instead, switch to MULTIXACT_MEMBER_AUTOVAC_THRESHOLD (eq 2^32) members
threshold. It is used to determine if we need to force autovacuum or not.
Author: Maxim Orlov <[email protected]>
---
src/backend/access/transam/multixact.c | 117 +++----------------------
src/backend/commands/vacuum.c | 2 +-
src/backend/postmaster/autovacuum.c | 4 +-
src/include/access/multixact.h | 1 -
4 files changed, 15 insertions(+), 109 deletions(-)
diff --git a/src/backend/access/transam/multixact.c b/src/backend/access/transam/multixact.c
index c51e03e832..c1f228c5fb 100644
--- a/src/backend/access/transam/multixact.c
+++ b/src/backend/access/transam/multixact.c
@@ -204,10 +204,14 @@ MXOffsetToMemberOffset(MultiXactOffset offset)
member_in_group * sizeof(TransactionId);
}
-/* Multixact members wraparound thresholds. */
-#define MULTIXACT_MEMBER_SAFE_THRESHOLD (MaxMultiXactOffset / 2)
-#define MULTIXACT_MEMBER_DANGER_THRESHOLD \
- (MaxMultiXactOffset - MaxMultiXactOffset / 4)
+/*
+ * Multixact members warning threshold.
+ *
+ * If difference bettween nextOffset and oldestOffset exceed this value, we
+ * trigger autovacuumin order to release the disk space, reduce table bloat if
+ * possible.
+ */
+#define MULTIXACT_MEMBER_AUTOVAC_THRESHOLD UINT64CONST(0xFFFFFFFF)
static inline MultiXactId
PreviousMultiXactId(MultiXactId multi)
@@ -2616,15 +2620,13 @@ GetOldestMultiXactId(void)
}
/*
- * Determine how aggressively we need to vacuum in order to prevent member
- * wraparound.
+ * Determine if we need to vacuum for member or not.
*
* To do so determine what's the oldest member offset and install the limit
* info in MultiXactState, where it can be used to prevent overrun of old data
* in the members SLRU area.
*
- * The return value is true if emergency autovacuum is required and false
- * otherwise.
+ * The return value is true if autovacuum is required and false otherwise.
*/
static bool
SetOffsetVacuumLimit(bool is_startup)
@@ -2712,10 +2714,10 @@ SetOffsetVacuumLimit(bool is_startup)
LWLockRelease(MultiXactGenLock);
/*
- * Do we need an emergency autovacuum? If we're not sure, assume yes.
+ * Do we need autovacuum? If we're not sure, assume yes.
*/
return !oldestOffsetKnown ||
- (nextOffset - oldestOffset > MULTIXACT_MEMBER_SAFE_THRESHOLD);
+ (nextOffset - oldestOffset > MULTIXACT_MEMBER_AUTOVAC_THRESHOLD);
}
/*
@@ -2761,101 +2763,6 @@ find_multixact_start(MultiXactId multi, MultiXactOffset *result)
return true;
}
-/*
- * Determine how many multixacts, and how many multixact members, currently
- * exist. Return false if unable to determine.
- */
-static bool
-ReadMultiXactCounts(uint32 *multixacts, MultiXactOffset *members)
-{
- MultiXactOffset nextOffset;
- MultiXactOffset oldestOffset;
- MultiXactId oldestMultiXactId;
- MultiXactId nextMultiXactId;
- bool oldestOffsetKnown;
-
- LWLockAcquire(MultiXactGenLock, LW_SHARED);
- nextOffset = MultiXactState->nextOffset;
- oldestMultiXactId = MultiXactState->oldestMultiXactId;
- nextMultiXactId = MultiXactState->nextMXact;
- oldestOffset = MultiXactState->oldestOffset;
- oldestOffsetKnown = MultiXactState->oldestOffsetKnown;
- LWLockRelease(MultiXactGenLock);
-
- if (!oldestOffsetKnown)
- return false;
-
- *members = nextOffset - oldestOffset;
- *multixacts = nextMultiXactId - oldestMultiXactId;
- return true;
-}
-
-/*
- * Multixact members can be removed once the multixacts that refer to them
- * are older than every datminmxid. autovacuum_multixact_freeze_max_age and
- * vacuum_multixact_freeze_table_age work together to make sure we never have
- * too many multixacts; we hope that, at least under normal circumstances,
- * this will also be sufficient to keep us from using too many offsets.
- * However, if the average multixact has many members, we might exhaust the
- * members space while still using few enough members that these limits fail
- * to trigger relminmxid advancement by VACUUM. At that point, we'd have no
- * choice but to start failing multixact-creating operations with an error.
- *
- * To prevent that, if more than a threshold portion of the members space is
- * used, we effectively reduce autovacuum_multixact_freeze_max_age and
- * to a value just less than the number of multixacts in use. We hope that
- * this will quickly trigger autovacuuming on the table or tables with the
- * oldest relminmxid, thus allowing datminmxid values to advance and removing
- * some members.
- *
- * As the fraction of the member space currently in use grows, we become
- * more aggressive in clamping this value. That not only causes autovacuum
- * to ramp up, but also makes any manual vacuums the user issues more
- * aggressive. This happens because vacuum_get_cutoffs() will clamp the
- * freeze table and the minimum freeze age cutoffs based on the effective
- * autovacuum_multixact_freeze_max_age this function returns. In the worst
- * case, we'll claim the freeze_max_age to zero, and every vacuum of any
- * table will freeze every multixact.
- */
-int
-MultiXactMemberFreezeThreshold(void)
-{
- MultiXactOffset members;
- uint32 multixacts;
- uint32 victim_multixacts;
- double fraction;
- int result;
-
- /* If we can't determine member space utilization, assume the worst. */
- if (!ReadMultiXactCounts(&multixacts, &members))
- return 0;
-
- /* If member space utilization is low, no special action is required. */
- if (members <= MULTIXACT_MEMBER_SAFE_THRESHOLD)
- return autovacuum_multixact_freeze_max_age;
-
- /*
- * Compute a target for relminmxid advancement. The number of multixacts
- * we try to eliminate from the system is based on how far we are past
- * MULTIXACT_MEMBER_SAFE_THRESHOLD.
- */
- fraction = (double) (members - MULTIXACT_MEMBER_SAFE_THRESHOLD);
- fraction /= (double) (MULTIXACT_MEMBER_DANGER_THRESHOLD - MULTIXACT_MEMBER_SAFE_THRESHOLD);
-
- victim_multixacts = multixacts * fraction;
-
- /* fraction could be > 1.0, but lowest possible freeze age is zero */
- if (victim_multixacts > multixacts)
- return 0;
- result = multixacts - victim_multixacts;
-
- /*
- * Clamp to autovacuum_multixact_freeze_max_age, so that we never make
- * autovacuum less aggressive than it would otherwise be.
- */
- return Min(result, autovacuum_multixact_freeze_max_age);
-}
-
typedef struct mxtruncinfo
{
int64 earliestExistingPage;
diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c
index 86f36b3695..e7506e268a 100644
--- a/src/backend/commands/vacuum.c
+++ b/src/backend/commands/vacuum.c
@@ -1133,7 +1133,7 @@ vacuum_get_cutoffs(Relation rel, const VacuumParams *params,
* normally autovacuum_multixact_freeze_max_age, but may be less if we are
* short of multixact member space.
*/
- effective_multixact_freeze_max_age = MultiXactMemberFreezeThreshold();
+ effective_multixact_freeze_max_age = autovacuum_multixact_freeze_max_age;
/*
* Almost ready to set freeze output parameters; check if OldestXmin or
diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index dc3cf87aba..180bb7e96e 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -1122,7 +1122,7 @@ do_start_worker(void)
/* Also determine the oldest datminmxid we will consider. */
recentMulti = ReadNextMultiXactId();
- multiForceLimit = recentMulti - MultiXactMemberFreezeThreshold();
+ multiForceLimit = recentMulti - autovacuum_multixact_freeze_max_age;
if (multiForceLimit < FirstMultiXactId)
multiForceLimit -= FirstMultiXactId;
@@ -1915,7 +1915,7 @@ do_autovacuum(void)
* normally autovacuum_multixact_freeze_max_age, but may be less if we are
* short of multixact member space.
*/
- effective_multixact_freeze_max_age = MultiXactMemberFreezeThreshold();
+ effective_multixact_freeze_max_age = autovacuum_multixact_freeze_max_age;
/*
* Find the pg_database entry and select the default freeze ages. We use
diff --git a/src/include/access/multixact.h b/src/include/access/multixact.h
index 90583634ec..5aefbddce3 100644
--- a/src/include/access/multixact.h
+++ b/src/include/access/multixact.h
@@ -143,7 +143,6 @@ extern void MultiXactSetNextMXact(MultiXactId nextMulti,
extern void MultiXactAdvanceNextMXact(MultiXactId minMulti,
MultiXactOffset minMultiOffset);
extern void MultiXactAdvanceOldest(MultiXactId oldestMulti, Oid oldestMultiDB);
-extern int MultiXactMemberFreezeThreshold(void);
extern void multixact_twophase_recover(TransactionId xid, uint16 info,
void *recdata, uint32 len);
--
2.43.0
[application/octet-stream] v8-0003-Make-pg_upgrade-convert-multixact-offsets.patch (18.3K, 10-v8-0003-Make-pg_upgrade-convert-multixact-offsets.patch)
download | inline diff:
From c324315152346d7f2090aaf79b142726aa2486ae Mon Sep 17 00:00:00 2001
From: Maxim Orlov <[email protected]>
Date: Tue, 13 Aug 2024 14:44:50 +0300
Subject: [PATCH v8 3/5] Make pg_upgrade convert multixact offsets.
Author: Maxim Orlov <[email protected]>
Author: Yura Sokolov <[email protected]>
---
src/bin/pg_upgrade/Makefile | 1 +
src/bin/pg_upgrade/meson.build | 1 +
src/bin/pg_upgrade/pg_upgrade.c | 42 ++-
src/bin/pg_upgrade/pg_upgrade.h | 14 +-
src/bin/pg_upgrade/segresize.c | 541 ++++++++++++++++++++++++++++++++
5 files changed, 594 insertions(+), 5 deletions(-)
create mode 100644 src/bin/pg_upgrade/segresize.c
diff --git a/src/bin/pg_upgrade/Makefile b/src/bin/pg_upgrade/Makefile
index f83d2b5d30..70908d63a3 100644
--- a/src/bin/pg_upgrade/Makefile
+++ b/src/bin/pg_upgrade/Makefile
@@ -21,6 +21,7 @@ OBJS = \
info.o \
option.o \
parallel.o \
+ segresize.o \
pg_upgrade.o \
relfilenumber.o \
server.o \
diff --git a/src/bin/pg_upgrade/meson.build b/src/bin/pg_upgrade/meson.build
index 3d88419674..16f898ba14 100644
--- a/src/bin/pg_upgrade/meson.build
+++ b/src/bin/pg_upgrade/meson.build
@@ -10,6 +10,7 @@ pg_upgrade_sources = files(
'info.c',
'option.c',
'parallel.c',
+ 'segresize.c',
'pg_upgrade.c',
'relfilenumber.c',
'server.c',
diff --git a/src/bin/pg_upgrade/pg_upgrade.c b/src/bin/pg_upgrade/pg_upgrade.c
index 663235816f..1654e877c0 100644
--- a/src/bin/pg_upgrade/pg_upgrade.c
+++ b/src/bin/pg_upgrade/pg_upgrade.c
@@ -750,8 +750,42 @@ copy_xact_xlog_xid(void)
if (old_cluster.controldata.cat_ver >= MULTIXACT_FORMATCHANGE_CAT_VER &&
new_cluster.controldata.cat_ver >= MULTIXACT_FORMATCHANGE_CAT_VER)
{
- copy_subdir_files("pg_multixact/offsets", "pg_multixact/offsets");
- copy_subdir_files("pg_multixact/members", "pg_multixact/members");
+ /*
+ * If the old server is before the MULTIXACTOFFSET_FORMATCHANGE_CAT_VER
+ * it must have 32-bit multixid offsets, thus it should be converted.
+ */
+ if (old_cluster.controldata.cat_ver < MULTIXACTOFFSET_FORMATCHANGE_CAT_VER &&
+ new_cluster.controldata.cat_ver >= MULTIXACTOFFSET_FORMATCHANGE_CAT_VER)
+ {
+ MultiXactOffset oldest_offset,
+ next_offset;
+
+ remove_new_subdir("pg_multixact/offsets", false);
+ prep_status("Converting pg_multixact/offsets to 64-bit");
+ oldest_offset = convert_multixact_offsets();
+ check_ok();
+
+ remove_new_subdir("pg_multixact/members", false);
+ prep_status("Converting pg_multixact/members");
+ convert_multixact_members(oldest_offset);
+ check_ok();
+
+ next_offset = old_cluster.controldata.chkpnt_nxtmxoff;
+ if (oldest_offset)
+ {
+ if (next_offset < oldest_offset)
+ next_offset += ((MultiXactOffset) 1 << 32) - 1;
+
+ next_offset -= oldest_offset - 1;
+
+ old_cluster.controldata.chkpnt_nxtmxoff = next_offset;
+ }
+ }
+ else
+ {
+ copy_subdir_files("pg_multixact/offsets", "pg_multixact/offsets");
+ copy_subdir_files("pg_multixact/members", "pg_multixact/members");
+ }
prep_status("Setting next multixact ID and offset for new cluster");
@@ -760,9 +794,9 @@ copy_xact_xlog_xid(void)
* counters here and the oldest multi present on system.
*/
exec_prog(UTILITY_LOG_FILE, NULL, true, true,
- "\"%s/pg_resetwal\" -O %u -m %u,%u \"%s\"",
+ "\"%s/pg_resetwal\" -O %llu -m %u,%u \"%s\"",
new_cluster.bindir,
- old_cluster.controldata.chkpnt_nxtmxoff,
+ (unsigned long long) old_cluster.controldata.chkpnt_nxtmxoff,
old_cluster.controldata.chkpnt_nxtmulti,
old_cluster.controldata.chkpnt_oldstMulti,
new_cluster.pgdata);
diff --git a/src/bin/pg_upgrade/pg_upgrade.h b/src/bin/pg_upgrade/pg_upgrade.h
index 53f693c2d4..2c85ec1e94 100644
--- a/src/bin/pg_upgrade/pg_upgrade.h
+++ b/src/bin/pg_upgrade/pg_upgrade.h
@@ -114,6 +114,13 @@ extern char *output_files[];
*/
#define MULTIXACT_FORMATCHANGE_CAT_VER 201301231
+/*
+ * Swicth from 32-bit to 64-bit for multixid offsets.
+ *
+ * XXX: should be changed to the actual CATALOG_VERSION_NO on commit.
+ */
+#define MULTIXACTOFFSET_FORMATCHANGE_CAT_VER 202409041
+
/*
* large object chunk size added to pg_controldata,
* commit 5f93c37805e7485488480916b4585e098d3cc883
@@ -230,7 +237,7 @@ typedef struct
uint32 chkpnt_nxtepoch;
uint32 chkpnt_nxtoid;
uint32 chkpnt_nxtmulti;
- uint32 chkpnt_nxtmxoff;
+ uint64 chkpnt_nxtmxoff;
uint32 chkpnt_oldstMulti;
uint32 chkpnt_oldstxid;
uint32 align;
@@ -515,3 +522,8 @@ typedef struct
FILE *file;
char path[MAXPGPATH];
} UpgradeTaskReport;
+
+/* segresize.c */
+
+MultiXactOffset convert_multixact_offsets(void);
+void convert_multixact_members(MultiXactOffset oldest_offset);
diff --git a/src/bin/pg_upgrade/segresize.c b/src/bin/pg_upgrade/segresize.c
new file mode 100644
index 0000000000..2f6f3b3288
--- /dev/null
+++ b/src/bin/pg_upgrade/segresize.c
@@ -0,0 +1,541 @@
+/*
+ * segresize.c
+ *
+ * SLRU segment resize utility
+ *
+ * Copyright (c) 2024, PostgreSQL Global Development Group
+ * src/bin/pg_upgrade/segresize.c
+ */
+
+#include "postgres_fe.h"
+
+#include "pg_upgrade.h"
+#include "access/multixact.h"
+
+/* See slru.h */
+#define SLRU_PAGES_PER_SEGMENT 32
+
+/*
+ * Some kind of iterator associated with a particular SLRU segment. The idea is
+ * to specify the segment and page number and then move through the pages.
+ */
+typedef struct SlruSegState
+{
+ char *dir;
+ char *fn;
+ FILE *file;
+ int64 segno;
+ uint64 pageno;
+ bool leading_gap;
+ bool long_segment_names;
+} SlruSegState;
+
+/*
+ * Mirrors the SlruFileName from slru.c
+ */
+static inline char *
+SlruFileName(SlruSegState *state)
+{
+ if (state->long_segment_names)
+ {
+ Assert(state->segno >= 0 &&
+ state->segno <= INT64CONST(0xFFFFFFFFFFFFFFF));
+ return psprintf("%s/%015llX", state->dir, (long long) state->segno);
+ }
+ else
+ {
+ Assert(state->segno >= 0 && state->segno <= INT64CONST(0xFFFFFF));
+ return psprintf("%s/%04X", state->dir, (unsigned int) state->segno);
+ }
+}
+
+/*
+ * Create new SLRU segment file.
+ */
+static void
+create_segment(SlruSegState *state)
+{
+ Assert(state->fn == NULL);
+ Assert(state->file == NULL);
+
+ state->fn = SlruFileName(state);
+ state->file = fopen(state->fn, "wb");
+ if (!state->file)
+ pg_fatal("could not create file \"%s\": %m", state->fn);
+}
+
+/*
+ * Open existing SLRU segment file.
+ */
+static void
+open_segment(SlruSegState *state)
+{
+ Assert(state->fn == NULL);
+ Assert(state->file == NULL);
+
+ state->fn = SlruFileName(state);
+ state->file = fopen(state->fn, "rb");
+ if (!state->file)
+ pg_fatal("could not open file \"%s\": %m", state->fn);
+}
+
+/*
+ * Close SLRU segment file.
+ */
+static void
+close_segment(SlruSegState *state)
+{
+ if (state->file)
+ {
+ fclose(state->file);
+ state->file = NULL;
+ }
+
+ if (state->fn)
+ {
+ pfree(state->fn);
+ state->fn = NULL;
+ }
+}
+
+/*
+ * Read next page from the old 32-bit offset segment file.
+ */
+static int
+read_old_segment_page(SlruSegState *state, void *buf, bool *empty)
+{
+ int len;
+
+ /* Open next segment file, if needed. */
+ if (!state->fn)
+ {
+ if (!state->segno)
+ state->leading_gap = true;
+
+ open_segment(state);
+
+ /* Set position to the needed page. */
+ if (state->pageno > 0 &&
+ fseek(state->file, state->pageno * BLCKSZ, SEEK_SET))
+ {
+ close_segment(state);
+ }
+ }
+
+ if (state->file)
+ {
+ /* Segment file do exists, read page from it. */
+ state->leading_gap = false;
+
+ len = fread(buf, sizeof(char), BLCKSZ, state->file);
+
+ /* Are we done or was there an error? */
+ if (len <= 0)
+ {
+ if (ferror(state->file))
+ pg_fatal("error reading file \"%s\": %m", state->fn);
+
+ if (feof(state->file))
+ {
+ *empty = true;
+ len = -1;
+
+ close_segment(state);
+ }
+ }
+ else
+ *empty = false;
+ }
+ else if (!state->leading_gap)
+ {
+ /* We reached the last segment. */
+ len = -1;
+ *empty = true;
+ }
+ else
+ {
+ /* Skip few first segments if they were frozen and removed. */
+ len = BLCKSZ;
+ *empty = true;
+ }
+
+ if (++state->pageno >= SLRU_PAGES_PER_SEGMENT)
+ {
+ /* Start a new segment. */
+ state->segno++;
+ state->pageno = 0;
+
+ close_segment(state);
+ }
+
+ return len;
+}
+
+/*
+ * Write next page to the new 64-bit offset segment file.
+ */
+static void
+write_new_segment_page(SlruSegState *state, void *buf)
+{
+ /*
+ * Create a new segment file if we still didn't. Creation is
+ * postponed until the first non-empty page is found. This helps
+ * not to create completely empty segments.
+ */
+ if (!state->file)
+ {
+ create_segment(state);
+
+ /* Write zeroes to the previously skipped prefix. */
+ if (state->pageno > 0)
+ {
+ char zerobuf[BLCKSZ] = {0};
+
+ for (int64 i = 0; i < state->pageno; i++)
+ {
+ if (fwrite(zerobuf, sizeof(char), BLCKSZ, state->file) != BLCKSZ)
+ pg_fatal("could not write file \"%s\": %m", state->fn);
+ }
+ }
+ }
+
+ /* Write page to the new segment (if it was created). */
+ if (state->file)
+ {
+ if (fwrite(buf, sizeof(char), BLCKSZ, state->file) != BLCKSZ)
+ pg_fatal("could not write file \"%s\": %m", state->fn);
+ }
+
+ /*
+ * Did we reach the maximum page number? Then close segment file
+ * and create a new one on the next iteration.
+ */
+ if (++state->pageno >= SLRU_PAGES_PER_SEGMENT)
+ {
+ /* Start a new segment. */
+ state->segno++;
+ state->pageno = 0;
+
+ close_segment(state);
+ }
+}
+
+typedef uint32 MultiXactOffsetOld;
+
+#define MaxMultiXactOffsetOld ((MultiXactOffsetOld) 0xFFFFFFFF)
+
+#define MULTIXACT_OFFSETS_PER_PAGE_OLD (BLCKSZ / sizeof(MultiXactOffsetOld))
+#define MULTIXACT_OFFSETS_PER_PAGE_NEW (BLCKSZ / sizeof(MultiXactOffset))
+
+/*
+ * Convert pg_multixact/offsets segments and return oldest multi offset.
+ */
+MultiXactOffset
+convert_multixact_offsets(void)
+{
+ SlruSegState oldseg = {0},
+ newseg = {0};
+ MultiXactOffsetOld oldbuf[MULTIXACT_OFFSETS_PER_PAGE_OLD] = {0};
+ MultiXactOffset newbuf[MULTIXACT_OFFSETS_PER_PAGE_NEW] = {0},
+ oldest_offset = 0;
+ uint64 oldest_multi = old_cluster.controldata.chkpnt_oldstMulti,
+ next_multi = old_cluster.controldata.chkpnt_nxtmulti,
+ multi,
+ old_entry,
+ new_entry;
+ bool oldest_offset_known = false;
+
+ oldseg.dir = psprintf("%s/pg_multixact/offsets", old_cluster.pgdata);
+ newseg.dir = psprintf("%s/pg_multixact/offsets", new_cluster.pgdata);
+
+ old_entry = oldest_multi % MULTIXACT_OFFSETS_PER_PAGE_OLD;
+ oldseg.pageno = oldest_multi / MULTIXACT_OFFSETS_PER_PAGE_OLD;
+ oldseg.segno = oldseg.pageno / SLRU_PAGES_PER_SEGMENT;
+ oldseg.pageno %= SLRU_PAGES_PER_SEGMENT;
+ oldseg.long_segment_names = false;
+
+ new_entry = oldest_multi % MULTIXACT_OFFSETS_PER_PAGE_NEW;
+ newseg.pageno = oldest_multi / MULTIXACT_OFFSETS_PER_PAGE_NEW;
+ newseg.segno = newseg.pageno / SLRU_PAGES_PER_SEGMENT;
+ newseg.pageno %= SLRU_PAGES_PER_SEGMENT;
+ newseg.long_segment_names = true;
+
+ if (next_multi < oldest_multi)
+ next_multi += (uint64) 1 << 32; /* wraparound */
+
+ /* Copy multi offsets reading only needed segment pages */
+ for (multi = oldest_multi; multi < next_multi; old_entry = 0)
+ {
+ int oldlen;
+ bool empty;
+
+ /* Handle possible segment wraparound */
+#define OLD_OFFSET_SEGNO_MAX \
+ (MaxMultiXactId / MULTIXACT_OFFSETS_PER_PAGE_OLD / SLRU_PAGES_PER_SEGMENT)
+ if (oldseg.segno > OLD_OFFSET_SEGNO_MAX)
+ {
+ oldseg.segno = 0;
+ oldseg.pageno = 0;
+ }
+
+ oldlen = read_old_segment_page(&oldseg, oldbuf, &empty);
+ if (empty || oldlen != BLCKSZ)
+ pg_fatal("cannot read page %llu from file \"%s\": %m",
+ (unsigned long long) oldseg.pageno, oldseg.fn);
+
+ /* Save oldest multi offset */
+ if (!oldest_offset_known)
+ {
+ oldest_offset = oldbuf[old_entry];
+ oldest_offset_known = true;
+ }
+
+ /* Skip wrapped-around invalid MultiXactIds */
+ if (multi == (uint64) 1 << 32)
+ {
+ Assert(oldseg.segno == 0);
+ Assert(oldseg.pageno == 1);
+ Assert(old_entry == 0);
+ Assert(new_entry == 0);
+
+ multi += FirstMultiXactId;
+ old_entry = FirstMultiXactId;
+ new_entry = FirstMultiXactId;
+ }
+
+ /* Copy entries to the new page */
+ for (; multi < next_multi && old_entry < MULTIXACT_OFFSETS_PER_PAGE_OLD;
+ multi++, old_entry++)
+ {
+ MultiXactOffset offset = oldbuf[old_entry];
+
+ /* Handle possible offset wraparound (1 becomes 2^32) */
+ if (offset < oldest_offset)
+ offset += ((uint64) 1 << 32) - 1;
+
+ /* Subtract oldest_offset, so new offsets will start from 1 */
+ newbuf[new_entry++] = offset - oldest_offset + 1;
+
+ if (new_entry >= MULTIXACT_OFFSETS_PER_PAGE_NEW)
+ {
+ /* Handle possible segment wraparound */
+#define NEW_OFFSET_SEGNO_MAX \
+ (MaxMultiXactId / MULTIXACT_OFFSETS_PER_PAGE_NEW / SLRU_PAGES_PER_SEGMENT)
+ if (newseg.segno > NEW_OFFSET_SEGNO_MAX)
+ {
+ newseg.segno = 0;
+ newseg.pageno = 0;
+ }
+
+ /* Write new page */
+ write_new_segment_page(&newseg, newbuf);
+ new_entry = 0;
+ }
+ }
+ }
+
+ /* Write the last incomplete page */
+ if (new_entry > 0 || oldest_multi == next_multi)
+ {
+ memset(&newbuf[new_entry], 0,
+ sizeof(newbuf[0]) * (MULTIXACT_OFFSETS_PER_PAGE_NEW - new_entry));
+ write_new_segment_page(&newseg, newbuf);
+ }
+
+ /* Use next_offset as oldest_offset, if oldest_multi == next_multi */
+ if (!oldest_offset_known)
+ {
+ Assert(oldest_multi == next_multi);
+ oldest_offset = (MultiXactOffset) old_cluster.controldata.chkpnt_nxtmxoff;
+ }
+
+ /* Release resources */
+ close_segment(&oldseg);
+ close_segment(&newseg);
+
+ pfree(oldseg.dir);
+ pfree(newseg.dir);
+
+ return oldest_offset;
+}
+
+#define MXACT_MEMBERS_FLAG_BYTES 1
+
+#define MULTIXACT_MEMBERS_PER_GROUP 4
+#define MULTIXACT_MEMBERGROUP_SIZE \
+ (MULTIXACT_MEMBERS_PER_GROUP * (sizeof(TransactionId) + MXACT_MEMBERS_FLAG_BYTES))
+#define MULTIXACT_MEMBERGROUPS_PER_PAGE \
+ (BLCKSZ / MULTIXACT_MEMBERGROUP_SIZE)
+
+#define MULTIXACT_MEMBERS_PER_PAGE \
+ (MULTIXACT_MEMBERS_PER_GROUP * MULTIXACT_MEMBERGROUPS_PER_PAGE)
+#define MULTIXACT_MEMBER_FLAG_BYTES_PER_GROUP \
+ (MXACT_MEMBERS_FLAG_BYTES * MULTIXACT_MEMBERS_PER_GROUP)
+
+typedef struct MultiXactMembersCtx
+{
+ SlruSegState seg;
+ char buf[BLCKSZ];
+ int group;
+ int member;
+ char *flag;
+ TransactionId *xid;
+} MultiXactMembersCtx;
+
+static void
+MultiXactMembersCtxInit(MultiXactMembersCtx *ctx)
+{
+ ctx->seg.dir = psprintf("%s/pg_multixact/members", new_cluster.pgdata);
+ ctx->seg.long_segment_names = false;
+
+ ctx->group = 0;
+ ctx->member = 1; /* skip invalid zero offset */
+
+ ctx->flag = (char *) ctx->buf + ctx->group * MULTIXACT_MEMBERGROUP_SIZE;
+ ctx->xid = (TransactionId *)(ctx->flag + MXACT_MEMBERS_FLAG_BYTES * MULTIXACT_MEMBERS_PER_GROUP);
+
+ ctx->flag += ctx->member;
+ ctx->xid += ctx->member;
+}
+
+static void
+MultiXactMembersCtxAdd(MultiXactMembersCtx *ctx, char flag, TransactionId xid)
+{
+ /* Copy member's xid and flags to the new page */
+ *ctx->flag++ = flag;
+ *ctx->xid++ = xid;
+
+ if (++ctx->member < MULTIXACT_MEMBERS_PER_GROUP)
+ return;
+
+ /* Start next member group */
+ ctx->member = 0;
+
+ if (++ctx->group >= MULTIXACT_MEMBERGROUPS_PER_PAGE)
+ {
+ /* Write current page and start new */
+ write_new_segment_page(&ctx->seg, ctx->buf);
+
+ ctx->group = 0;
+ memset(ctx->buf, 0, BLCKSZ);
+ }
+
+ ctx->flag = (char *) ctx->buf + ctx->group * MULTIXACT_MEMBERGROUP_SIZE;
+ ctx->xid = (TransactionId *)(ctx->flag + MXACT_MEMBERS_FLAG_BYTES * MULTIXACT_MEMBERS_PER_GROUP);
+}
+
+static void
+MultiXactMembersCtxFinit(MultiXactMembersCtx *ctx)
+{
+ if (ctx->flag > (char *) ctx->buf)
+ write_new_segment_page(&ctx->seg, ctx->buf);
+
+ close_segment(&ctx->seg);
+
+ pfree(ctx->seg.dir);
+}
+
+/*
+ * Convert pg_multixact/members segments, offsets will start from 1.
+ *
+ */
+void
+convert_multixact_members(MultiXactOffset oldest_offset)
+{
+ MultiXactOffset next_offset,
+ offset;
+ SlruSegState oldseg = {0};
+ char oldbuf[BLCKSZ] = {0};
+ int oldidx;
+ MultiXactMembersCtx newctx = {0};
+
+ oldseg.dir = psprintf("%s/pg_multixact/members", old_cluster.pgdata);
+
+ next_offset = (MultiXactOffset) old_cluster.controldata.chkpnt_nxtmxoff;
+ if (next_offset < oldest_offset)
+ next_offset += ((uint64) 1 << 32) - 1;
+
+ /* Initialize the old starting position */
+ oldseg.pageno = oldest_offset / MULTIXACT_MEMBERS_PER_PAGE;
+ oldseg.segno = oldseg.pageno / SLRU_PAGES_PER_SEGMENT;
+ oldseg.pageno %= SLRU_PAGES_PER_SEGMENT;
+ oldseg.long_segment_names = false;
+
+ /* Initialize new starting position */
+ MultiXactMembersCtxInit(&newctx);
+
+ /* Iterate through the original directory */
+ oldidx = oldest_offset % MULTIXACT_MEMBERS_PER_PAGE;
+ for (offset = oldest_offset; offset < next_offset;)
+ {
+ bool empty;
+ int oldlen;
+ int ngroups;
+ int oldgroup;
+ int oldmember;
+
+ oldlen = read_old_segment_page(&oldseg, oldbuf, &empty);
+ if (empty || oldlen != BLCKSZ)
+ pg_fatal("cannot read page %llu from file \"%s\": %m",
+ (unsigned long long) oldseg.pageno, oldseg.fn);
+
+ /* Iterate through the old member groups */
+ ngroups = oldlen / MULTIXACT_MEMBERGROUP_SIZE;
+ oldmember = oldidx % MULTIXACT_MEMBERS_PER_GROUP;
+ oldgroup = oldidx / MULTIXACT_MEMBERS_PER_GROUP;
+ while (oldgroup < ngroups && offset < next_offset)
+ {
+ char *oldflag;
+ TransactionId *oldxid;
+ int i;
+
+ oldflag = (char *) oldbuf + oldgroup * MULTIXACT_MEMBERGROUP_SIZE;
+ oldxid = (TransactionId *)(oldflag + MULTIXACT_MEMBER_FLAG_BYTES_PER_GROUP);
+
+ oldxid += oldmember;
+ oldflag += oldmember;
+
+ /* Iterate through the old members */
+ for (i = oldmember;
+ i < MULTIXACT_MEMBERS_PER_GROUP && offset < next_offset;
+ i++)
+ {
+ MultiXactMembersCtxAdd(&newctx, *oldflag++, *oldxid++);
+
+ if (++offset == (uint64) 1 << 32)
+ {
+ Assert(i == MaxMultiXactOffsetOld % MULTIXACT_MEMBERS_PER_GROUP);
+ goto wraparound;
+ }
+ }
+
+ oldgroup++;
+ oldmember = 0;
+ }
+
+ oldidx = 0;
+
+ continue;
+
+wraparound:
+#define SEGNO_MAX MaxMultiXactOffsetOld / MULTIXACT_MEMBERS_PER_PAGE / SLRU_PAGES_PER_SEGMENT
+#define PAGENO_MAX MaxMultiXactOffsetOld / MULTIXACT_MEMBERS_PER_PAGE % SLRU_PAGES_PER_SEGMENT
+ Assert((oldseg.segno == SEGNO_MAX && oldseg.pageno == PAGENO_MAX + 1) ||
+ (oldseg.segno == SEGNO_MAX + 1 && oldseg.pageno == 0));
+
+ /* Switch to segment 0000 */
+ close_segment(&oldseg);
+ oldseg.segno = 0;
+ oldseg.pageno = 0;
+
+ /* skip invalid zero multi offset */
+ oldidx = 1;
+ }
+
+ MultiXactMembersCtxFinit(&newctx);
+
+ /* Release resources */
+ close_segment(&oldseg);
+
+ pfree(oldseg.dir);
+}
--
2.43.0
view thread (21+ messages) latest in thread
reply
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: [email protected]
Cc: [email protected], [email protected], [email protected], [email protected], [email protected]
Subject: Re: POC: make mxidoff 64 bits
In-Reply-To: <CACG=ezZGQFBb0yepka8hU2BmJ48ujt3xa+aYLNL0BQPx0vqwZg@mail.gmail.com>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox