Horton

Re: pgsql: Add test for postmaster crash restarts.

Andres Freund <andres@anarazel.de>

View thread View on postgr.es
From: Andres Freund <andres@anarazel.de>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: pgsql-committers@postgresql.org
Date: 2017-09-19T17:42:32Z
Lists: pgsql-hackers 
On 2017-09-19 09:58:26 -0700, Andres Freund wrote:
> 
> 
> On September 19, 2017 9:53:28 AM PDT, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> >Well, please fix it ASAP, if you don't want to take it out pending
> >the fixes.
> 
> Will as soon as I finished my morning coffee. Uncaffeinated, which my phone fittingly autocorrects to unvaccinated, commits aren't a good idea.

Done. Survived ~100 cycles here locally, while running make -j16 -s
check-world in two parallel branches. But I have a fast disk, so it's
not comparable.

If the buildfarm doesn't complain about the use of IPC::Run's timeout
functionality, we should probably patch that into the other use of
IPC::Run as well, but especially into the other user of the pump() until
... scheme.

Greetings,

Andres Freund

Commits

  1. Allow pg_ctl kill to send SIGKILL.

  2. Try to make crash restart test work on windows.

  3. Make new crash restart test a bit more robust.

  4. Add test for postmaster crash restarts.