agora inbox for [email protected]
help / color / mirror / Atom feedRe: raising the default default_statistics_target
506+ messages / 5 participants
[nested] [flat]
* Re: raising the default default_statistics_target
@ 2004-03-08 17:17 Josh Berkus <[email protected]>
2004-03-08 18:08 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
0 siblings, 1 reply; 506+ messages in thread
From: Josh Berkus @ 2004-03-08 17:17 UTC (permalink / raw)
To: pgsql-hackers; +Cc: Neil Conway <[email protected]>
Neil,
> In the simple test I performed, raising the default_statistics_target
> from 10 to 25 resulted in a 40% increase in the time to ANALYZE a
> large table. (I picked 25 more or less at random -- would 15 or 20 be
> better?)
I find that very interesting, since I haven't found much higher increases to
be a proportionate penality. For example, on an 11-column table raising 3
columns to statistics=250 merely doubled the ANALYZE time. I have not done
exact timing, but would be happy to ....
>It may also be the case that for those people for whom 10
> is an insufficient stats target, 25 is also insufficient.
It is. I've found that "problem" queries, especially those caused by real,
uneven distribution of data, require raising statistics to 150-400 in order
to fix. This is much to high a level to assign as a default.
> Any comments on whether increasing the default stats target is a good
> idea for 7.5? (Details on the test I performed are included below)
No. I don't think it's a good idea to raise the default for *all* columns;
for one thing, I'd really hate to think what, say, a default stats of 100
would do to a TEXT column with an average of 8K of data per row.
Further, in 7.5 we'll be introducing correlated stats for multi-column indexes
(unless something's gone off with that?) which should help a lot of problem
queries. And change our whole emphasis on brute forcing analyze through
increasing stats into the 100's.
If you really want to tackle this issue, though, here's what I suggest:
1) add a GUC called default_statistics_indexed, which starts at say 100 or 50.
2) When ever the user indexes a column, automatically increase the stats
to the level in default_statistics_indexed, if they are at the level in
default_statistics_target.
This will then give indexed columns "automatically" a somewhat higher level of
stats analysis than other columns. This should help a lot of "slow query"
problems, yet effectively leave the selection of "important" columns in the
hands of the DBA. Make sense?
Also, another great feature in this department would be to extend the
multi-column correlation statistics to cover foriegn keys, as a way of
improving cross-table estimates.
Anyway, keep me in the loop on this, I have a lot of very complex databases I
can test such issues on.
--
-Josh Berkus
Aglio Database Solutions
San Francisco
^ permalink raw reply [nested|flat] 506+ messages in thread
* Re: raising the default default_statistics_target
2004-03-08 17:17 Re: raising the default default_statistics_target Josh Berkus <[email protected]>
@ 2004-03-08 18:08 ` Tom Lane <[email protected]>
2004-03-08 18:28 ` Re: raising the default default_statistics_target Bruce Momjian <[email protected]>
2004-03-08 19:07 ` Re: raising the default default_statistics_target Josh Berkus <[email protected]>
0 siblings, 2 replies; 506+ messages in thread
From: Tom Lane @ 2004-03-08 18:08 UTC (permalink / raw)
To: [email protected]; +Cc: pgsql-hackers; Neil Conway <[email protected]>
Josh Berkus <[email protected]> writes:
> It is. I've found that "problem" queries, especially those caused by real,
> uneven distribution of data, require raising statistics to 150-400 in order
> to fix. This is much to high a level to assign as a default.
That's basically what's bothering me about the suggestion to increase to
25 --- I'm dubious that it will do any good.
> Further, in 7.5 we'll be introducing correlated stats for multi-column indexes
> (unless something's gone off with that?)
News to me. It's certainly not there now.
> This will then give indexed columns "automatically" a somewhat higher
> level of stats analysis than other columns.
That is potentially a good idea. There's still the question of what is
a reasonable default, though.
regards, tom lane
^ permalink raw reply [nested|flat] 506+ messages in thread
* Re: raising the default default_statistics_target
2004-03-08 17:17 Re: raising the default default_statistics_target Josh Berkus <[email protected]>
2004-03-08 18:08 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
@ 2004-03-08 18:28 ` Bruce Momjian <[email protected]>
2004-03-08 18:37 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
1 sibling, 1 reply; 506+ messages in thread
From: Bruce Momjian @ 2004-03-08 18:28 UTC (permalink / raw)
To: Tom Lane <[email protected]>; +Cc: [email protected]; pgsql-hackers; Neil Conway <[email protected]>
Tom Lane wrote:
> Josh Berkus <[email protected]> writes:
> > It is. I've found that "problem" queries, especially those caused by real,
> > uneven distribution of data, require raising statistics to 150-400 in order
> > to fix. This is much to high a level to assign as a default.
>
> That's basically what's bothering me about the suggestion to increase to
> 25 --- I'm dubious that it will do any good.
>
> > Further, in 7.5 we'll be introducing correlated stats for multi-column indexes
> > (unless something's gone off with that?)
>
> News to me. It's certainly not there now.
>
> > This will then give indexed columns "automatically" a somewhat higher
> > level of stats analysis than other columns.
>
> That is potentially a good idea. There's still the question of what is
> a reasonable default, though.
Do all the columns have to have the same number of statistics buckets?
Could that stats collector adjust the number of buckets based on the
data somehow?
--
Bruce Momjian | http://candle.pha.pa.us
[email protected] | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073
^ permalink raw reply [nested|flat] 506+ messages in thread
* Re: raising the default default_statistics_target
2004-03-08 17:17 Re: raising the default default_statistics_target Josh Berkus <[email protected]>
2004-03-08 18:08 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
2004-03-08 18:28 ` Re: raising the default default_statistics_target Bruce Momjian <[email protected]>
@ 2004-03-08 18:37 ` Tom Lane <[email protected]>
2004-03-08 19:22 ` Re: raising the default default_statistics_target Bruce Momjian <[email protected]>
0 siblings, 1 reply; 506+ messages in thread
From: Tom Lane @ 2004-03-08 18:37 UTC (permalink / raw)
To: Bruce Momjian <[email protected]>; +Cc: [email protected]; pgsql-hackers; Neil Conway <[email protected]>
Bruce Momjian <[email protected]> writes:
> Do all the columns have to have the same number of statistics buckets?
They do not, but the effort spent by ANALYZE is proportional to the
largest stats target among all the columns of the table.
regards, tom lane
^ permalink raw reply [nested|flat] 506+ messages in thread
* Re: raising the default default_statistics_target
2004-03-08 17:17 Re: raising the default default_statistics_target Josh Berkus <[email protected]>
2004-03-08 18:08 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
2004-03-08 18:28 ` Re: raising the default default_statistics_target Bruce Momjian <[email protected]>
2004-03-08 18:37 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
@ 2004-03-08 19:22 ` Bruce Momjian <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Bruce Momjian @ 2004-03-08 19:22 UTC (permalink / raw)
To: Tom Lane <[email protected]>; +Cc: [email protected]; pgsql-hackers; Neil Conway <[email protected]>
Tom Lane wrote:
> Bruce Momjian <[email protected]> writes:
> > Do all the columns have to have the same number of statistics buckets?
>
> They do not, but the effort spent by ANALYZE is proportional to the
> largest stats target among all the columns of the table.
Could we use previous stats to determine how many buckets to use when
running ANALYZE. Also, if columns have a different number of buckets,
does that mean that we don't have the same per-query overhead for a
larger stats target?
--
Bruce Momjian | http://candle.pha.pa.us
[email protected] | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073
^ permalink raw reply [nested|flat] 506+ messages in thread
* Re: raising the default default_statistics_target
2004-03-08 17:17 Re: raising the default default_statistics_target Josh Berkus <[email protected]>
2004-03-08 18:08 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
@ 2004-03-08 19:07 ` Josh Berkus <[email protected]>
2004-03-08 19:19 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
1 sibling, 1 reply; 506+ messages in thread
From: Josh Berkus @ 2004-03-08 19:07 UTC (permalink / raw)
To: Tom Lane <[email protected]>; +Cc: pgsql-hackers; Neil Conway <[email protected]>
Tom,
> > Further, in 7.5 we'll be introducing correlated stats for multi-column
indexes
> > (unless something's gone off with that?)
This was discussed on Hackers in October, a complete implementation was shown,
I thought it was committed at that time. If not, what happened to it?
Dammit, it's impossible to find anything in the archives if you don't have
some good keywords or at least the author. Is the autor reading this? Will
you speak up?
--
-Josh Berkus
Aglio Database Solutions
San Francisco
^ permalink raw reply [nested|flat] 506+ messages in thread
* Re: raising the default default_statistics_target
2004-03-08 17:17 Re: raising the default default_statistics_target Josh Berkus <[email protected]>
2004-03-08 18:08 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
2004-03-08 19:07 ` Re: raising the default default_statistics_target Josh Berkus <[email protected]>
@ 2004-03-08 19:19 ` Tom Lane <[email protected]>
2004-03-08 19:41 ` Re: raising the default default_statistics_target Josh Berkus <[email protected]>
0 siblings, 1 reply; 506+ messages in thread
From: Tom Lane @ 2004-03-08 19:19 UTC (permalink / raw)
To: [email protected]; +Cc: pgsql-hackers; Neil Conway <[email protected]>
Josh Berkus <[email protected]> writes:
> Further, in 7.5 we'll be introducing correlated stats for multi-column
> indexes (unless something's gone off with that?)
> This was discussed on Hackers in October, a complete implementation was shown,
> I thought it was committed at that time. If not, what happened to it?
Are you sure you're not thinking of stats for functional indexes?
regards, tom lane
^ permalink raw reply [nested|flat] 506+ messages in thread
* Re: raising the default default_statistics_target
2004-03-08 17:17 Re: raising the default default_statistics_target Josh Berkus <[email protected]>
2004-03-08 18:08 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
2004-03-08 19:07 ` Re: raising the default default_statistics_target Josh Berkus <[email protected]>
2004-03-08 19:19 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
@ 2004-03-08 19:41 ` Josh Berkus <[email protected]>
2004-03-09 14:33 ` Re: raising the default default_statistics_target Robert Treat <[email protected]>
0 siblings, 1 reply; 506+ messages in thread
From: Josh Berkus @ 2004-03-08 19:41 UTC (permalink / raw)
To: Tom Lane <[email protected]>; +Cc: pgsql-hackers; Neil Conway <[email protected]>
Tom,
> Are you sure you're not thinking of stats for functional indexes?
Positive. I even remember seeing that the patch was accepted.
The patch specifically had to do with a multi-column correlation algorithm for
improving the selectivity of multi-column indexes.
Problem is, with 1400 posts per month August to October, I can't find it, and
the keywords that I think are obvious don't turn anything up.
--
-Josh Berkus
Aglio Database Solutions
San Francisco
^ permalink raw reply [nested|flat] 506+ messages in thread
* Re: raising the default default_statistics_target
2004-03-08 17:17 Re: raising the default default_statistics_target Josh Berkus <[email protected]>
2004-03-08 18:08 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
2004-03-08 19:07 ` Re: raising the default default_statistics_target Josh Berkus <[email protected]>
2004-03-08 19:19 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
2004-03-08 19:41 ` Re: raising the default default_statistics_target Josh Berkus <[email protected]>
@ 2004-03-09 14:33 ` Robert Treat <[email protected]>
2004-03-09 15:11 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
0 siblings, 1 reply; 506+ messages in thread
From: Robert Treat @ 2004-03-09 14:33 UTC (permalink / raw)
To: Josh Berkus <[email protected]>; +Cc: Tom Lane <[email protected]>; pgsql-hackers; Neil Conway <[email protected]>
I think the thread you're thinking of is on or about this post:
http://archives.postgresql.org/pgsql-patches/2003-12/msg00039.php
Manfred Koizar produced a patch that modified index correlation by
sorting equal key values based on item pointers. The patch went as far
as getting accepted into the patch queue, but Tom raised some doubts
about it and it was subsequently removed.
Robert Treat
On Mon, 2004-03-08 at 14:41, Josh Berkus wrote:
> Tom,
>
> > Are you sure you're not thinking of stats for functional indexes?
>
> Positive. I even remember seeing that the patch was accepted.
>
> The patch specifically had to do with a multi-column correlation algorithm for
> improving the selectivity of multi-column indexes.
>
> Problem is, with 1400 posts per month August to October, I can't find it, and
> the keywords that I think are obvious don't turn anything up.
>
> --
> -Josh Berkus
> Aglio Database Solutions
> San Francisco
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 2: you can get off all lists at once with the unregister command
> (send "unregister YourEmailAddressHere" to [email protected])
--
Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL
^ permalink raw reply [nested|flat] 506+ messages in thread
* Re: raising the default default_statistics_target
2004-03-08 17:17 Re: raising the default default_statistics_target Josh Berkus <[email protected]>
2004-03-08 18:08 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
2004-03-08 19:07 ` Re: raising the default default_statistics_target Josh Berkus <[email protected]>
2004-03-08 19:19 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
2004-03-08 19:41 ` Re: raising the default default_statistics_target Josh Berkus <[email protected]>
2004-03-09 14:33 ` Re: raising the default default_statistics_target Robert Treat <[email protected]>
@ 2004-03-09 15:11 ` Tom Lane <[email protected]>
2004-03-09 16:41 ` Re: raising the default default_statistics_target Josh Berkus <[email protected]>
0 siblings, 1 reply; 506+ messages in thread
From: Tom Lane @ 2004-03-09 15:11 UTC (permalink / raw)
To: Robert Treat <[email protected]>; +Cc: Josh Berkus <[email protected]>; pgsql-hackers; Neil Conway <[email protected]>
Robert Treat <[email protected]> writes:
> I think the thread you're thinking of is on or about this post:
> http://archives.postgresql.org/pgsql-patches/2003-12/msg00039.php
> Manfred Koizar produced a patch that modified index correlation by
> sorting equal key values based on item pointers. The patch went as far
> as getting accepted into the patch queue, but Tom raised some doubts
> about it and it was subsequently removed.
Hm, that had nothing to do with multi-column correlation though.
I'm at a loss to think of any work that matches with Josh's
recollection.
regards, tom lane
^ permalink raw reply [nested|flat] 506+ messages in thread
* Re: raising the default default_statistics_target
2004-03-08 17:17 Re: raising the default default_statistics_target Josh Berkus <[email protected]>
2004-03-08 18:08 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
2004-03-08 19:07 ` Re: raising the default default_statistics_target Josh Berkus <[email protected]>
2004-03-08 19:19 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
2004-03-08 19:41 ` Re: raising the default default_statistics_target Josh Berkus <[email protected]>
2004-03-09 14:33 ` Re: raising the default default_statistics_target Robert Treat <[email protected]>
2004-03-09 15:11 ` Re: raising the default default_statistics_target Tom Lane <[email protected]>
@ 2004-03-09 16:41 ` Josh Berkus <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Josh Berkus @ 2004-03-09 16:41 UTC (permalink / raw)
To: Tom Lane <[email protected]>; Robert Treat <[email protected]>; +Cc: pgsql-hackers; Neil Conway <[email protected]>
Guys,
> Hm, that had nothing to do with multi-column correlation though.
> I'm at a loss to think of any work that matches with Josh's
> recollection.
Hmmmm .... it's possible that early e-mails about Manfred's patch claimed to
improve performance for multi-column indexes.
But it's also possible I'm remembering something else.
Darn it, though! 'cause multi-column correlation is one of our big issues on
estimates for complex queries.
--
-Josh Berkus
Aglio Database Solutions
San Francisco
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
* [PATCH] Restructure repack worker teardown
@ 2026-05-18 17:13 Álvaro Herrera <[email protected]>
0 siblings, 0 replies; 506+ messages in thread
From: Álvaro Herrera @ 2026-05-18 17:13 UTC (permalink / raw)
The original code would leave a shared memory segment unreleased if we
fail partway through initialization. Change the shutdown order so that
we already free it.
Author: Álvaro Herrera <[email protected]>
Discussion: https://postgr.es/m/[email protected]
---
src/backend/commands/repack.c | 67 ++++++++++++++++-------------------
1 file changed, 31 insertions(+), 36 deletions(-)
diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c
index bfc62c8f752..c9064d8fd13 100644
--- a/src/backend/commands/repack.c
+++ b/src/backend/commands/repack.c
@@ -3411,10 +3411,14 @@ start_repack_decoding_worker(Oid relid)
shm_mq_handle *mqh;
BackgroundWorker bgw;
+ decoding_worker = palloc0_object(DecodingWorker);
+
/* Setup shared memory. */
size = BUFFERALIGN(offsetof(DecodingWorkerShared, error_queue)) +
BUFFERALIGN(REPACK_ERROR_QUEUE_SIZE);
seg = dsm_create(size, 0);
+ decoding_worker->seg = seg;
+
shared = (DecodingWorkerShared *) dsm_segment_address(seg);
shared->initialized = false;
shared->lsn_upto = InvalidXLogRecPtr;
@@ -3454,14 +3458,12 @@ start_repack_decoding_worker(Oid relid)
bgw.bgw_main_arg = UInt32GetDatum(dsm_segment_handle(seg));
bgw.bgw_notify_pid = MyProcPid;
- decoding_worker = palloc0_object(DecodingWorker);
if (!RegisterDynamicBackgroundWorker(&bgw, &decoding_worker->handle))
ereport(ERROR,
errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED),
errmsg("out of background worker slots"),
errhint("You might need to increase \"%s\".", "max_worker_processes"));
- decoding_worker->seg = seg;
decoding_worker->error_mqh = mqh;
/*
@@ -3487,17 +3489,6 @@ start_repack_decoding_worker(Oid relid)
ConditionVariableCancelSleep();
}
-/*
- * PG_ENSURE_ERROR_CLEANUP callback to stop the decoding worker.
- * This ensures the worker is terminated on both ERROR and FATAL exits,
- * unlike PG_FINALLY which only handles ERROR.
- */
-static void
-repack_decoding_worker_cleanup_cb(int code, Datum arg)
-{
- stop_repack_decoding_worker();
-}
-
/*
* Stop the decoding worker and cleanup the related resources.
*
@@ -3508,39 +3499,43 @@ static void
stop_repack_decoding_worker(void)
{
BgwHandleStatus status;
+ dsm_segment *dsmseg;
- /* Haven't reached the worker startup? */
+ /* Nothing to do if no worker was set up. */
if (decoding_worker == NULL)
return;
- /* Could not register the worker? */
- if (decoding_worker->handle == NULL)
- return;
+ /* Terminate the worker process, if one is running. */
+ if (decoding_worker->handle != NULL)
+ {
+ TerminateBackgroundWorker(decoding_worker->handle);
+ /* The worker should really exit before the REPACK command does. */
+ HOLD_INTERRUPTS();
+ status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
+ RESUME_INTERRUPTS();
- TerminateBackgroundWorker(decoding_worker->handle);
- /* The worker should really exit before the REPACK command does. */
- HOLD_INTERRUPTS();
- status = WaitForBackgroundWorkerShutdown(decoding_worker->handle);
- RESUME_INTERRUPTS();
-
- if (status == BGWH_POSTMASTER_DIED)
- ereport(FATAL,
- errcode(ERRCODE_ADMIN_SHUTDOWN),
- errmsg("postmaster exited during REPACK command"));
-
- shm_mq_detach(decoding_worker->error_mqh);
+ if (status == BGWH_POSTMASTER_DIED)
+ ereport(FATAL,
+ errcode(ERRCODE_ADMIN_SHUTDOWN),
+ errmsg("postmaster exited during REPACK command"));
+ }
/*
- * If we could not cancel the current sleep due to ERROR, do that before
- * we detach from the shared memory the condition variable is located in.
- * If we did not, the bgworker ERROR handling code would try and fail
- * badly.
+ * Now detach from our shared memory segment. In error cases there might
+ * still be messages from the worker in the queue, which ProcessInterrupts
+ * would try to read; this is pointless (and causes an assertion failure),
+ * so set the global pointer to NULL to have ProcessRepackMessages ignore
+ * them.
*/
- ConditionVariableCancelSleep();
-
- dsm_detach(decoding_worker->seg);
+ dsmseg = decoding_worker->seg;
pfree(decoding_worker);
decoding_worker = NULL;
+
+ /* We must also cancel the current sleep, if one is still set up */
+ ConditionVariableCancelSleep();
+
+ if (dsmseg != NULL)
+ dsm_detach(dsmseg);
}
/* stop_repack_decoding_worker, wrapped as a before_shmem_exit callback */
--
2.47.3
--b42ct6adeyjj4cbm--
^ permalink raw reply [nested|flat] 506+ messages in thread
end of thread, other threads:[~2026-05-18 17:13 UTC | newest]
Thread overview: 506+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2004-03-08 17:17 Re: raising the default default_statistics_target Josh Berkus <[email protected]>
2004-03-08 18:08 ` Tom Lane <[email protected]>
2004-03-08 18:28 ` Bruce Momjian <[email protected]>
2004-03-08 18:37 ` Tom Lane <[email protected]>
2004-03-08 19:22 ` Bruce Momjian <[email protected]>
2004-03-08 19:07 ` Josh Berkus <[email protected]>
2004-03-08 19:19 ` Tom Lane <[email protected]>
2004-03-08 19:41 ` Josh Berkus <[email protected]>
2004-03-09 14:33 ` Robert Treat <[email protected]>
2004-03-09 15:11 ` Tom Lane <[email protected]>
2004-03-09 16:41 ` Josh Berkus <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[email protected]>
2026-05-18 17:13 [PATCH] Restructure repack worker teardown Álvaro Herrera <[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