Hi,
On 2023-02-14 03:25:04 +0000, PG Bug reporting form wrote:
> The following bug has been logged on the website:
>
> Bug reference: 17791
> Logged by: Robins Tharakan
> Email address: tharakan@gmail.com
> PostgreSQL version: 12.14
> Operating system: Ubuntu 20.04
> Description:
>
> This assert() is:
> - Not easily reproducible + rare (~twice a month)
> - Always surfaces only on REL_12_STABLE (never on v13+)
> - The line numbers changed (in Nov 22) but IIUC its the same signature
> - SQLs seem irrelevant. Can dig up full SQLs, if required.
> - I provide 4 backtraces, 2 SELECTs, 1 UPDATE, 1 DELETE.
>
>
> Posting here because I noticed the above pattern today, and thought
> someone could identify what's so specifically wrong with v12. Note that
> the bug could pre-date Oct 22, since I wasn't aggressively testing older
> versions before then.
Are there any log messages around the crashes? I'm wondering whether we're
erroring out while aborting, to then abort again, or such. Which will often
elicit warning/error messages.
Unfortunately, without further information, I doubt this is sufficient for
anybody to debug the issue. The stack traces all seem to be during error
handling, after catching an error in PostgresMain(). So there's really not
much to go on.
Greetings,
Andres Freund