Horton

Re: stress test for parallel workers

Tom Lane <tgl@sss.pgh.pa.us>

View thread View on postgr.es
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Justin Pryzby <pryzby@telsasoft.com>
Cc: Thomas Munro <thomas.munro@gmail.com>, pgsql-hackers <pgsql-hackers@postgresql.org>
Date: 2019-07-23T23:29:04Z
Lists: pgsql-hackers 
Justin Pryzby <pryzby@telsasoft.com> writes:
> I want to say I'm almost certain it wasn't ENOSPC in other cases, since,
> failing to find log output, I ran df right after the failure.

The fact that you're not finding log output matching what was reported
to the client seems to me to be a mighty strong indication that there
*was* an ENOSPC problem.  Can you reconfigure to put the postmaster
log on a different volume?

> But that gives me an idea: is it possible there's an issue with files being
> held opened by worker processes ?  Including by parallel workers?  Probably
> WALs, even after they're rotated ?  If there were worker processes holding
> opened lots of rotated WALs, that could cause ENOSPC, but that wouldn't be
> obvious after they die, since the space would then be freed.

Parallel workers aren't ever allowed to write, in the current
implementation, so it's not real obvious why they'd have any
WAL log files open at all.

			regards, tom lane

Commits

  1. In the postmaster, rely on the signal infrastructure to block signals.

  2. Paper over regression failures in infinite_recurse() on PPC64 Linux.

  3. Hack pg_ctl to report postmaster's exit status.

  4. Re-order some regression test scripts for more parallelism.