On 2012-09-06, Kobus Wolvaardt <kobuswolf@gmail.com> wrote:
> Something that is curios is that if a DB takes long, it really takes
> horribly long like some kind of a lock is holding it. It would sit at a few
> kb dump size for 20 minutes en then run a bit and get stuck again (as far
> as we can tell), what we do know is that it is way to slow for some IO or
> cpu starvation. We have seen a 5GB backup sitting at 1Gb after 12hours and
> then we stop it.
> Any siggestions? Can autovacume or lack thereof cause this? It seems noone
> has been doing any maintenance on the DB (it does look like autovacuum is
> running), so any suggestions would be nice.
perhaps some sort of DML could cause this, I've seen DML hang during
dumps
turn on logging of slow queries and check the logs after the next
speed disaster.
--
⚂⚃ 100% natural