Re: how to investigate GIN fast updates and cleanup cycles?

On Fri, Aug 28, 2015 at 12:10 PM Jeff Janes <jeff(dot)janes(at)gmail(dot)com> wrote:

> Did you change the system-wide autovacuum_analyze_scale_factor? If so,
> don't do that. You can use a table's storage parameters to set a custom
> autovacuum_analyze_scale_factor just for individual tables. So just the
> table with the troublesome gin index:
>

No I did it just to the problematic table:

ALTER TABLE my_table SET (autovacuum_vacuum_scale_factor=0);
ALTER TABLE my_table SET (autovacuum_analyze_scale_factor=0);

My system-wide autovacuum_analyze_threshold is 50 so I think that's ok.

There is a bulk load going on right now so a lot of tables are needing
vacuuming. I really need to increase my autovacuum_max_workers.

>
> Killing existing vacuums won't help (at least not until change your system
> wide setting back and so the above instead) because it will likely just
> restart on the same table it was on when you killed it.
>

You're right, I gave up trying, and am back running a manual VACUUM. It's
slow, but I can't bounce the db right now.