Hi,
On 2023-04-11 13:35:38 -0700, Andres Freund wrote:
> On 2023-04-11 14:46:23 -0500, Justin Pryzby wrote:
> > Yes, $SUBJECT is correct.
> >
> > On an old centos6 VM which I'd forgotten about and never removed from
> > monitoring, I noticed that a process had recently crashed...
> >
> > Maybe this is an issue which was already fixed, but I looked and find no
> > bug report nor patch about it. Feel free to dismiss the problem report
> > if it's not interesting or useful.
>
> > postgres was compiled locally at 4e54d231a. It'd been running
> > continuously since September without crashing until a couple weeks ago
> > (and running nearly-continuously for months before that).
>
> It possibly could be:
>
> Author: Andres Freund <andres@anarazel.de>
> Branch: master [cb2e7ddfe] 2022-12-02 18:10:30 -0800
> Branch: REL_15_STABLE Release: REL_15_2 [c6a60471a] 2022-12-02 18:07:47 -0800
> Branch: REL_14_STABLE Release: REL_14_7 [6344bc097] 2022-12-02 18:10:30 -0800
> Branch: REL_13_STABLE Release: REL_13_10 [7944d2d8c] 2022-12-02 18:13:40 -0800
> Branch: REL_12_STABLE Release: REL_12_14 [35b99a18f] 2022-12-02 18:16:14 -0800
> Branch: REL_11_STABLE Release: REL_11_19 [af3517c15] 2022-12-02 18:17:54 -0800
>
> Prevent pgstats from getting confused when relkind of a relation changes
>
> But the fact that it's on a catalog table's stats makes it less likely,
> although not impossible.
>
>
> Any chance there were conversions from tables to views in that connection?
Nope, not possible - the stack trace actually shows this is during connection establishment.
Thomas, see stack trace upthread?
Greetings,
Andres Freund