Re: autovacuum process blocks without reporting a deadlock

On Nov 27, 2007 4:52 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> You didn't happen to note what 9293 was doing did you?  It's living
> fairly dangerously in any case by trying to acquire exclusive lock
> when it already holds a bunch of other lower-level locks; that's a
> recipe for deadlock if I ever saw one.
>
>                         regards, tom lane
>

Ah ok, 9293 is a triggerd process and tries to "ALTER TABLE ...
DISABLE TRIGGER (other trigger)" and so implicitly tries to acquire an
AccessExclusiveLock and runs in a deadlock?

So is better not to use "ALTER TABLE ... " in triggerfunctions,
because there are always existing lower-level locks?

regards,
thomas