| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
|---|---|
| To: | Alvaro Herrera <alvherre(at)commandprompt(dot)com> |
| Cc: | Ian Westmacott <ianw(at)intellivid(dot)com>, pgsql-admin(at)postgresql(dot)org |
| Subject: | Re: 8.1.8 autovacuum missing databases |
| Date: | 2008-04-30 16:27:49 |
| Message-ID: | 19778.1209572869@sss.pgh.pa.us |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-admin |
Alvaro Herrera <alvherre(at)commandprompt(dot)com> writes:
> Right -- if you have two databases over the wraparound edge, it will
> alternate between them.
> I still think that the autovac is dying before completing the task. Did
> you investigate whether there are "ERROR" messages coming from
> autovacuum? No PG crashes would happen.
AFAICS, if it's triggering a whole_db run and the run fails to get to
the point of updating pg_database, then the next run will pick the very
same db.
I'm noticing though that the 8.1 logic pays attention to both
datvacuumxid and datfrozenxid. Could we see the age() of both of
those columns for all the databases?
regards, tom lane
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Ian Westmacott | 2008-04-30 17:01:48 | Re: 8.1.8 autovacuum missing databases |
| Previous Message | Alvaro Herrera | 2008-04-30 16:19:42 | Re: 8.1.8 autovacuum missing databases |