Thread: postmaster freeze

Hello pgsql-general,

I have problem with my postgres server. I am running postgres
7.3.4 on RedHat9 (shrike) on dual processor machine.
Sometimes (cca 3 times per day) postmaster service will freeze
with 100% CPU load with SELECT statement (statement is never
finished). If I try the same action again without restarting
server, it works (but freezed postmasted still consume 100% CPU).
It seems that something in postgres is locked, because
if I run command vacuum when this situation happens,
psql will freeze too. I defined also statement_timeout=600000
(10 minutes) in config, but It seem does not work.

Did somebody have the same problem?
Please, help, this db is productive...

--
 Vladimir Drobny                          mailto:vladimir.drobny@vvexpert.sk

Vladimir Drobny <vladimir.drobny@vvexpert.sk> writes:
> I have problem with my postgres server. I am running postgres
> 7.3.4 on RedHat9 (shrike) on dual processor machine.
> Sometimes (cca 3 times per day) postmaster service will freeze
> with 100% CPU load with SELECT statement (statement is never
> finished). If I try the same action again without restarting
> server, it works (but freezed postmasted still consume 100% CPU).
> It seems that something in postgres is locked, because
> if I run command vacuum when this situation happens,
> psql will freeze too. I defined also statement_timeout=600000
> (10 minutes) in config, but It seem does not work.

Hmm, are you able to "kill -9" the stuck backend?  I'm wondering about
hardware or kernel-level problems.  If kill doesn't work then the
problem is kernel or hardware.

If it does, then please try attaching to the stuck backend with gdb to
get a stack trace.  That would give some clue what it's doing.

            regards, tom lane