Re: Weird failure with latches in curculio on v15
Andres Freund <andres@anarazel.de>
From: Andres Freund <andres@anarazel.de>
To: Michael Paquier <michael@paquier.xyz>
Cc: Postgres hackers <pgsql-hackers@lists.postgresql.org>
Date: 2023-02-01T02:12:06Z
Lists: pgsql-hackers
Hi,
On 2023-02-01 10:53:17 +0900, Michael Paquier wrote:
> While browsing the buildfarm, I have noticed this failure on curcilio:
> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=curculio&dt=2023-02-01%2001%3A05%3A17
>
> The test that has reported a failure is the check on the archive
> module callback:
> # Failed test 'check shutdown callback of shell archive module'
> # at t/020_archive_status.pl line 248.
> # Looks like you failed 1 test of 17.
> [02:28:06] t/020_archive_status.pl ..............
> Dubious, test returned 1 (wstat 256, 0x100)
> Failed 1/17 subtests
>
> Looking closer, this is a result of an assertion failure in the latch
> code:
> 2023-02-01 02:28:05.615 CET [6961:8] LOG: received fast shutdown request
> 2023-02-01 02:28:05.615 CET [6961:9] LOG: aborting any active transactions
> 2023-02-01 02:28:05.616 CET [30681:9] LOG: process 30681 releasing ProcSignal slot 33, but it contains 0
> TRAP: FailedAssertion("latch->owner_pid == MyProcPid", File: "latch.c", Line: 451, PID: 30681)
Given the ProcSignal LOG message before it, I don't think this is about
latches.
> The information available in standby2.log shows that 30681 is the
> startup process. I am not sure what all that means, yet.
>
> Thoughts or comments welcome.
Perhaps a wild write overwriting shared memory state?
Greetings,
Andres Freund
Commits
-
Avoid calling proc_exit() in processes forked by system().
- d0e7f95b4845 11.22 landed
- e2e16904224a 12.17 landed
- ac1dfc303d0e 13.13 landed
- 54fc9dca5b10 14.10 landed
- c9265ae80b6a 15.5 landed
- ee06199fcb0a 16.1 landed
- 97550c071197 17.0 landed
-
Move extra code out of the Pre/PostRestoreCommand() section.
- 882e522d6468 15.5 landed
- d1c56ad37b96 16.1 landed
- 8fb13dd6ab5b 17.0 landed
-
Revert refactoring of restore command code to shell_restore.c
- 2f6e15ac93c5 16.0 landed
-
Refactor code in charge of running shell-based recovery commands
- 9a740f81eb02 16.0 cited
-
Clean up inconsistent use of fflush().
- 7fed801135ba 16.0 cited
-
Report wait events for local shell commands like archive_command.
- 1b06d7bac901 15.0 cited