Horton

Re: [sqlsmith] Unpinning error in parallel worker

Jonathan Rudenberg <jonathan@titanous.com>

View thread View on postgr.es
From: Jonathan Rudenberg <jonathan@titanous.com>
To: Robert Haas <robertmhaas@gmail.com>, Thomas Munro <thomas.munro@enterprisedb.com>
Cc: Andreas Seltenreich <seltenreich@gmx.de>, Pg Hackers <pgsql-hackers@postgresql.org>
Date: 2018-04-17T20:52:31Z
Lists: pgsql-hackers 
On Wed, Mar 29, 2017, at 10:50, Robert Haas wrote:
> On Wed, Mar 29, 2017 at 1:31 AM, Thomas Munro
> <thomas.munro@enterprisedb.com> wrote:
> > I considered whether the error message could be improved but it
> > matches the message for an existing similar case (where you try to
> > attach to an unknown handle).
> 
> Ugh, OK.  I committed this, but I think this whole file needs a visit
> from the message style police.

I ran into an issue that matches the symptoms of this bug on a production cluster running 10.3 (Ubuntu 10.3-1.pgdg16.04+1) twice in the past week.

Hundreds of queries stuck with a wait_event of DynamicSharedMemoryControlLock and pg_terminate_backend did not terminate the queries.

In the log:

> FATAL:  cannot unpin a segment that is not pinned

I don't have a backtrace yet, but I will provide them if/when the issue happens again.

Jonathan

Commits

  1. Use English, instead of internal names, for translatable messages.

  2. Plug race in dsa_attach.