Alan McKay <alan.mckay@gmail.com> writes:
> I've got Munin installed on all my systems, so was able to get some
> interesting data around the big crash we had last night. We'd
> thought it was simply a matter of our DB connections maxing out, but
> it looks a bit more complex than that. A good 2 or 3 hours before the
> connections max (at 300), about 80% of them go to 'waiting'. I'm
> wondering what other data I should be looking at to see why.
pg_locks? Somebody taking exclusive lock on a widely-used table might
explain that.
regards, tom lane