Horton

Re: Weird failure with latches in curculio on v15

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: Andres Freund <andres@anarazel.de>
Cc: Tom Lane <tgl@sss.pgh.pa.us>, Robert Haas <robertmhaas@gmail.com>, Thomas Munro <thomas.munro@gmail.com>, Fujii Masao <fujii@postgresql.org>, Nathan Bossart <nathandbossart@gmail.com>, Postgres hackers <pgsql-hackers@lists.postgresql.org>
Date: 2023-02-02T01:23:21Z
Lists: pgsql-hackers 
On Wed, Feb 01, 2023 at 10:18:27AM -0800, Andres Freund wrote:
> On 2023-02-01 12:27:19 -0500, Tom Lane wrote:
>> Andres Freund <andres@anarazel.de> writes:
>> The main thing that system() brings to the table is platform-specific
>> knowledge of where the shell is.  I'm not very sure that we want to
>> wire in "/bin/sh".
> 
> We seem to be doing OK with using SHELLPROG in pg_regress, which just
> seems to be using $SHELL from the build environment.

It looks like this had better centralize a bit more of the logic from
pg_regress.c if that were to happen, to avoid more fuzzy logic with
WIN32.  That becomes invasive for a back-patch.

By the way, there is something that's itching me a bit here.  9a740f8
has enlarged by a lot the window between PreRestoreCommand() and
PostRestoreCommand(), however curculio has reported a failure on
REL_15_STABLE, where we only manipulate my_wait_event_info while the
flag is on.  Or I am getting that right that there is no way out of it
unless we remove the dependency to system() even in the back-branches?
Could there be an extra missing piece here?
--
Michael

Commits

  1. Avoid calling proc_exit() in processes forked by system().

  2. Move extra code out of the Pre/PostRestoreCommand() section.

  3. Revert refactoring of restore command code to shell_restore.c

  4. Refactor code in charge of running shell-based recovery commands

  5. Clean up inconsistent use of fflush().

  6. Report wait events for local shell commands like archive_command.