Re: Resetting recovery target parameters in pg_createsubscriber

On Fri, Jan 16, 2026 at 08:25:34AM +0900, Michael Paquier wrote:
> On Thu, Jan 15, 2026 at 07:05:20AM +0000, Hayato Kuroda (Fujitsu) wrote:
> > Thanks for working on the project. I found that one of BF animal fairywren has
> > been failing [1] the pg_createsubscriber test since the commit 639352d904c.
> > Can you help solving the issue?
>
> Thanks for the poke.  I'll look into it in details.

pg_ctl's 001_start_stop.pl, or even ssl_passphrase_callback's
001_testfunc.pl seem to be in line with what you are suggesting
Kuroda-san.  Both tests include an "external" pg_ctl start command and
explicitely stop the node to prevent an issue like the one we are
seeing here.  Hence, your suggestion of fix should be enough.  I have
added a comment to explain why a direct pg_ctl command is used, and
applied the result.

Let's now see if fairywren gets back to green.  I'll keep an eye on
the buildfarm.
--
Michael