On 2024-07-25 Th 12:00 AM, Alexander Lakhin wrote:
> Hello Andrew,
>
> 04.06.2024 13:00, Alexander Lakhin wrote:
>> Also, 027_stream_regress still fails due to the same reason:
>> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=serinus&dt=2024-05-22%2021%3A55%3A03
>>
>> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=flaviventris&dt=2024-05-22%2021%3A54%3A50
>>
>> (It's remarkable that these two animals failed at the same time.)
>>
>
> It looks like crake is failing now due to other reasons (not just
> concurrency) — it produced 10+ failures of the
> 027_stream_regress test starting from July, 9.
>
> The first such failure on REL_16_STABLE was [1], and that was the
> first run
> with 'PG_TEST_EXTRA' => '... wal_consistency_checking'.
>
> There is one known issue related to wal_consistency_checking [2], but I
> see no "incorrect resource manager data checksum" in the failure log...
>
> Moreover, the first such failure on REL_17_STABLE was [3], but that run
> was performed without wal_consistency_checking, as far as I can see.
>
> Can that failure be also related to the OS upgrade (I see that back in
> June crake was running on Fedora 39, but now it's running on Fedora 40)?
>
> So maybe we have two factors combined or there is another one?
>
> [1]
> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=crake&dt=2024-07-17%2014%3A56%3A09
> [2]
> https://www.postgresql.org/message-id/055bb33c-bccc-bc1d-c2f8-8598534448ac@gmail.com
> [3]
> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=crake&dt=2024-07-09%2021%3A37%3A04
Unlikely. The change in OS version was on June 17, more than a month ago.
But yes we do seem to have seen a lot of recovery_check failures on
crake in the last 8 days, which is roughly when I changed PG_TEST_EXTRA
to get more coverage.
cheers
andrew
--
Andrew Dunstan
EDB: https://www.enterprisedb.com