On Thu, 2006-03-30 at 11:10, Chris Travers wrote:
> lmyho wrote:
>
> >
> >The last one (successful bootup and shutdown):
> >pg_log# more postgresql-2006-03-30_103237.log
> >2006-03-30 10:32:37 CSTLOG: database system was shut down at 2006-03-30 10:32:20
> >CST
> >2006-03-30 10:32:37 CSTLOG: checkpoint record is at 0/33B96C
> >2006-03-30 10:32:37 CSTLOG: redo record is at 0/33B96C; undo record is at 0/0;
> >shutdown TRUE
> >2006-03-30 10:32:37 CSTLOG: next transaction ID: 5839; next OID: 16386
> >2006-03-30 10:32:37 CSTLOG: next MultiXactId: 1; next MultiXactOffset: 0
> >2006-03-30 10:32:37 CSTLOG: database system is ready
> >2006-03-30 10:32:37 CSTLOG: transaction ID wrap limit is 2147484146, limited by
> >database "postgres"
> >2006-03-30 10:32:37 CSTLOG: autovacuum: processing database "template1"
> >2006-03-30 10:33:37 CSTLOG: autovacuum: processing database "postgres"
> >2006-03-30 10:33:42 CSTLOG: incomplete startup packet
> >2006-03-30 10:33:42 CSTLOG: received fast shutdown request
> >2006-03-30 10:33:42 CSTLOG: shutting down
> >2006-03-30 10:33:42 CSTLOG: database system is shut down
> >2006-03-30 10:33:42 CSTLOG: logger shutting down
> >
> >You can see they all have that line! Can you see what's may be the problem? Thank
> >you!!!
> >
> >
> Interesting.... Database starts up at 10:32:37 and processes all
> databases through autovacuum.
> Then exactly a minute later, it processes the postgres database.
> 5 seconds later, you get a SIGINT.
>
> Is there something wrong with autovacuum? Normally I would doubt it but
> the other possibility is something is waiting for just over a minute and
> then shutting down the database.
>
> If you su postgres, can you run pg_ctl and get the database up by itself?
If I remember correctly, this is a problem with the debian startup
script requiring trust for the postgres superuser, and was discussed
sometime last year. I'll wander through the archives in a free moment
and see if I can find it.