From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Wayne Schroeder <raz(at)positivenetworks(dot)net> |
Cc: | pgsql-general(at)postgresql(dot)org |
Subject: | Re: Long running transactions |
Date: | 2006-03-21 00:45:27 |
Message-ID: | 9535.1142901927@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
Wayne Schroeder <raz(at)positivenetworks(dot)net> writes:
> I have a quite heavily accessed database that is vacuumed nightly.
> Attached to that database are processes, that at times, due to the way
> they are coded, stay "idle in transaction" for long periods of time
> (client lib implementation issues a BEGIN). I've been reading up on
> XIDs and the like and I'm trying to determine if this is going to cause
> problems.
Yup, it is; you should try to fix those apps to not send BEGIN until
they are ready to do some work. Idle-in-transaction processes will tend
to prevent VACUUM from removing dead rows, for instance, because it
can't be sure that those processes shouldn't be able to see recently-dead
rows.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Murali K. Maddali | 2006-03-21 02:16:03 | xml output. |
Previous Message | Wayne Schroeder | 2006-03-21 00:25:23 | Long running transactions |