Re: Replication slot drop message is sent after pgstats shutdown.

At Mon, 14 Feb 2022 17:20:16 -0800, Andres Freund <andres(at)anarazel(dot)de> wrote in
> Hi,
>
> On 2021-12-22 22:34:45 +0900, Masahiko Sawada wrote:
> > I've attached an updated patch. Please review it.
>
> Sorry for dropping the ball on this again :(. I've pushed the fix with some
> very minor polishing.

Thanks!

> > > The attached detects that bug, but I'm not sure it's worth expending
> > > test time, or this might be in the server test suit.
> >
> > Thanks. It's convenient to test this issue but I'm also not sure it's
> > worth adding to the test suit.
>
> I think it's definitely worth adding a test, but I don't particularly like the
> specific test implementation. Primarily because I think it's better to test
> this in a cluster that stays running, so that we can verify that the slot drop
> worked. It also doesn't seem necessary to create a separate cluster.

One of the points I was not satisfied the TAP test is the second point
above. FWIW I agree to the proposed test on the direction.

> I wrote the attached isolation test. I ended up not committing it yet - I'm
> worried that there could be some OS dependent output difference, due to some
> libpq error handling issues. See [1], which Tom pointed out is caused by the
> issue discussed in [2].

Mmm.. This is..
slot_creation_error.out
> step s2_init: <... completed>
> FATAL: terminating connection due to administrator command
> FATAL: terminating connection due to administrator command

> Greetings,
>
> Andres Freund
>
> [1] https://postgr.es/m/20220215004143.dlzsn72oqsmqa7uw%40alap3.anarazel.de
> [2] https://postgr.es/m/20220215004143.dlzsn72oqsmqa7uw%40alap3.anarazel.de

regards.

--
Kyotaro Horiguchi
NTT Open Source Software Center