Re: Tracking down a deadlock

Bill Moseley <moseley(at)hank(dot)org> writes:
> Not getting any nibbles, so allow me to try a short question:
> If I have a deadlock situation (that will be reported as such by
> Postgresql once the deadlock_timeout passes), does pg_stat_activity
> show the queries that are blocking each other?

In 8.2 or later it should do so; in prior versions there could be some
lag involved. Another thing to keep in mind is that backends like to
cache copies of the pg_stat_activity view --- if you are watching it
to see what is happening, your view is only current as of the start
of your current transaction. Or you can do pgstat_clear_snapshot()
to force collection of new info.

regards, tom lane