Dinesh Bhandary wrote:
> Hi Alvaro,
>
> I was killing to reproduce the bug.Once it is fixed we have no need to
> kill these jobs.
> Now we still have this production server not running in auto vacuum till
> we apply the fix during next maintenance window. Is running vacuum
> manually safe, or we may run into the same issue as auto vacuum tx
> wraparound job? I am thinking max_freeze set to 500 million, manual
> vacuum won¹t freeze xids, and it may be ok to run. But I am not completely
> sure and worried about the same issue happening in production even with
> manual vacuum. Please let me know.
Yes, that seems correct. I think you're only running into
for-wraparound vacuums because the relfrozenxid for that table is above
the 200 millions that IIRC are the default for max_freeze_age. You can
just set it to the maximum of 2000 million; the only downside is disk
space used by pg_clog and pg_multixact.
--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services