On 2020-Aug-15, Tom Lane wrote:
> hyrax's latest report suggests that this patch has issues under
> CLOBBER_CACHE_ALWAYS:
>
> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=hyrax&dt=2020-08-13%2005%3A09%3A58
>
> Hard to tell whether there's an actual bug there or just test instability,
> but either way it needs to be resolved.
Hmm, the only explanation I can see for this is that autovacuum managed
to summarize the range before the test script did it. So the solution
would simply be to disable autovacuum for the table across the whole
script.
I'm running the scripts and dependencies to verify that fix, but under
CLOBBER_CACHE_ALWAYS that takes quite a bit.
--
Álvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services