Thread: postmaster hangs
hi, i'm using postgres version 7.4.1 and it run very stable until, may be, 4 weeks ago. now the postmaster hangs sometimes for a while (actually since 2 hours) and doesn't accept connections but comes back anytime. there is no possibility to stop the postmaster. even a kill -9 (it's only the developement system) doesn't work. i have to reboot the system to get the database running again. actually ps brings: postgres 2148 1 0 Mar 11 ? 0:28 /usr/local/pgsql/bin/postmaster -i -d 2 top shows the postmaster process as follows: ? 2148 postgres 154 20 210M 244K run 0:28 0.02 0.02 postmaster the output of netstat is (grep for 5432): tcp 0 0 192.168.143.116.5432 192.168.143.116.61145 ESTABLISHED tcp 0 0 192.168.143.116.61145 192.168.143.116.5432 ESTABLISHED tcp 0 0 *.5432 *.* LISTEN tcp 0 0 192.168.143.116.5432 192.168.143.114.2550 ESTABLISHED 45d91540 stream 38 0 0 0 0 0 /tmp/.s.PGSQL.5432 45d916c0 stream 39 0 0 0 0 0 /tmp/.s.PGSQL.5432 40ed2680 stream 38 0 0 43191e80 0 0 /tmp/.s.PGSQL.5432 40ed2b00 stream 38 0 0 4458a400 0 0 /tmp/.s.PGSQL.5432 440c3b80 stream 37 0 0 0 0 0 /tmp/.s.PGSQL.5432 440c3d00 stream 37 0 0 0 0 0 /tmp/.s.PGSQL.5432 449e5100 stream 37 0 0 0 0 0 /tmp/.s.PGSQL.5432 43c65240 stream 37 0 0 4458a280 0 0 /tmp/.s.PGSQL.5432 43a15200 stream 0 0 43b48c40 0 0 0 /tmp/.s.PGSQL.5432 437db7c0 stream 37 0 0 0 0 0 /tmp/.s.PGSQL.5432 437dbc40 stream 37 0 0 0 0 0 /tmp/.s.PGSQL.5432 the server is a pa-risc system with a HP-UX 11.11, 64Bit does anybody has an idea, what could be the reason for it? thanks rolf
Rolf Staege <Rolf.Staege@med.uni-rostock.de> writes: > i'm using postgres version 7.4.1 and it run very stable until, may be, > 4 weeks ago. > now the postmaster hangs sometimes for a while (actually since 2 hours) > and doesn't accept connections but comes back anytime. there is no > possibility to stop the postmaster. even a kill -9 (it's only the > developement system) doesn't work. i have to reboot the system to get > the database running again. Hardware problems? Most of the cases I've heard of where kill -9 didn't work, it was because the process was partway through a disk operation and the kernel was holding off the signal until the I/O completed ... but there was something broken about the disk and so the operation never completed. In any case, there is *no* way for an unprivileged process like Postgres to resist kill -9, so your problem is certainly at either the hardware or kernel level. I'd suggest talking to HP ... regards, tom lane
On Tue, 2005-04-19 at 09:38, Rolf Staege wrote: > hi, > i'm using postgres version 7.4.1 and it run very stable until, may be, > 4 weeks ago. > now the postmaster hangs sometimes for a while (actually since 2 hours) > and doesn't accept connections but comes back anytime. there is no > possibility to stop the postmaster. even a kill -9 (it's only the > developement system) doesn't work. i have to reboot the system to get > the database running again. In addition to what Tom said (i.e. look at hardware problems) I'd also recommend updating the latest update of your versions of pgsql, 7.4.7.