On Thu, Oct 14, 2021 at 12:50 AM Masahiko Sawada wrote:
> Looking at the original commit, as you mentioned, ISTM performing
> pending list cleanup during (auto)analyze (and analyze_only) was
> introduced to perform the pending list cleanup on insert-only tables.
> Now that we have autovacuum_vacuu
On Thu, Oct 14, 2021 at 7:58 AM Peter Geoghegan wrote:
>
> On Sat, Oct 9, 2021 at 5:51 PM Peter Geoghegan wrote:
> > While I'm no closer to a backpatchable fix than I was on Thursday, I
> > do have some more ideas about what to do on HEAD. I now lean towards
> > completely ripping analyze_only ca
On Sat, Oct 9, 2021 at 5:51 PM Peter Geoghegan wrote:
> While I'm no closer to a backpatchable fix than I was on Thursday, I
> do have some more ideas about what to do on HEAD. I now lean towards
> completely ripping analyze_only calls out, there -- the whole idea of
> calling amvacuumcleanup() ro
On Thu, Oct 7, 2021 at 10:34 PM Peter Geoghegan wrote:
> This issue was brought to my attention by Nikolay Samokhvalov. He
> reached out privately about it. He mentioned one problematic case
> involving an ANALYZE lasting 45 minutes, or longer (per
> log_autovacuum_min_duration output for the auto
We generally only expect amvacuumcleanup() routines to be called
during VACUUM. But some ginvacuumcleanup() calls are an exception to
that general rule -- these are calls made during autoanalyze, where
ginvacuumcleanup() does real pending list cleanup work (not just a
no-op return). That'll only ha
