Horton

Re: pg_upgrade test failure

Thomas Munro <thomas.munro@gmail.com>

View thread View on postgr.es
From: Thomas Munro <thomas.munro@gmail.com>
To: Justin Pryzby <pryzby@telsasoft.com>
Cc: Andres Freund <andres@anarazel.de>, Michael Paquier <michael@paquier.xyz>, pgsql-hackers@lists.postgresql.org
Date: 2022-10-02T20:07:25Z
Lists: pgsql-hackers 
On Tue, Sep 20, 2022 at 1:31 PM Justin Pryzby <pryzby@telsasoft.com> wrote:
> I suspect that rmtree() was looping in pgunlink(), and got ENOENT, so
> didn't warn about the file itself, but then failed one moment later in
> rmdir.

Yeah, I think this is my fault.  In commit f357233c the new lstat()
call might return ENOENT for STATUS_DELETE_PENDING, and then we don't
enter pgunlink()'s 10 second sleep-retry loop.  Let me think about how
best to fix that, and how to write a regression test program that
would exercise stuff like this.  Might take a couple of days as I am
away from computers until mid-week.

Commits

  1. ci: freebsd: stop postgres instance on failure, to prevent log upload failures

  2. Try to fix pg_upgrade test on Windows, again.

  3. Refactor rmtree() to use get_dirent_type().

  4. Fix unlink() for STATUS_DELETE_PENDING on Windows.

  5. Move any remaining files generated by pg_upgrade into an internal subdir

  6. Make unlink() work for junction points on Windows.

  7. pg_upgrade: Move all the files generated internally to a subdirectory

  8. In pg_upgrade, try a few times to open a log file.