Hi,
On 2023-01-07 21:06:06 +0300, Michail Nikolaev wrote:
> 2) It is not an issue at table creation time. Issue is reproducible if
> vacuum_defer_cleanup_age set after table preparation.
>
> 3) To reproduce the issue, vacuum_defer_cleanup_age should flip xid
> over zero (be >= txid_current()).
> And it is stable.... So, for example - unable to reproduce with 733
> value, but 734 gives error each time.
> Just a single additional txid_current() (after data is filled) fixes a
> crash... It looks like the first SELECT FOR UPDATE + UPDATE silently
> poisons everything somehow.
> You could use such PSQL script:
FWIW, the concrete value for vacuum_defer_cleanup_age is not crucial to
encounter the problem. It needs to be a value that, when compared to the xid
that did the "INSERT INTO something_is_wrong_here", results in value <= 0.
Setting vacuum_defer_cleanup_age than the xid to a much larger value allows
the crash to be encountered repeatedly.
Greetings,
Andres Freund