Hi Tom,
> What you have not shown us is what transaction has
> actually *got* a lock on 16496.
Would you mind enlightening me as to how I can do so?
I stared at the output from pg_locks a long time and
just can't get more than what I had gotten out of.
Regards,
Tena Sakai
tsakai@gallo.ucsf.edu
-----Original Message-----
From: Tom Lane [mailto:tgl@sss.pgh.pa.us]
Sent: Mon 12/17/2007 8:25 PM
To: Tena Sakai
Cc: pgsql-admin@postgresql.org
Subject: Re: [ADMIN] How would I "close" a atble?
"Tena Sakai" <tsakai@gallo.ucsf.edu> writes:
> That pid 6697 caught my interest and I fished it out of "ps -ef":
> postgres 6697 4916 0 Dec15 ? 00:19:45 postgres: postgres canon [local] REINDEX waiting
> (and for the sake of completeness...)
> postgres 12930 4916 0 11:24 ? 00:00:00 postgres: gbrush canon 172.16.1.106(37819) SELECT waiting
> postgres 18177 4916 0 Dec15 ? 00:00:02 postgres: gbrush canon 172.16.1.106(53874) SELECT waiting
> postgres 18825 4916 0 12:55 ? 00:00:00 postgres: tsakai canon 127.0.0.1(44558) SELECT waiting
The deal here is that the REINDEX is blocked trying to get exclusive
lock on table 16496, while those other guys are (apparently) stacked
up behind it. What you have not shown us is what transaction has
actually *got* a lock on 16496.
> I could kill this process from unix
> level, but before I do so, would you please comment as to why
> this might have happened and what repercussion might I have
> from killing it, if any?
kill -INT would probably be safe enough, but you should first look into
what is blocking the REINDEX from going through.
regards, tom lane