On Wed, Feb 06, 2008 at 01:20:50PM +0530, Suresh Gupta VG wrote:
> I am on Solaris-9 and Pgsql-7.4.2. This database is running on the live
> production server. It takes more than 5 minutes to update one single
> query. This is very horrible and completely not acceptable on the live.
Take one of those slow queries and use "EXPLAIN ANALYZE" to figure out
why they are so slow.
BTW: Consider upgrading to the latest 7.4 (IIRC 7.4.11) - there should
be a lot of bugfixes you are missing, some of which may lead to data
corruption.
> We had scheduled Vacuum on daily basis at low traffic time of the day.
> Due to this, the client application is getting time out and the
> particular transaction is getting failed.
>
> We observed numbers of postmaster processes are running in the sleeping
> mode. We also observed some of the processes are running with the
> previous dates. The maximum connections configures is 100, but I can see
> number of transactions around 200 at some time of the day(May be peak
> time of transactions). Pls advice me to get rid of this.
>
> User Pid PPID Status Time Etime Stime
> Command
>
> -------- ---- ---- ------ --- ---------- --------
> -----------------
>
> pgsql 2519 1 S 1:45 30-20:50:51 Jan_06
> /db1/postgresql/bin/postmaster
This is just the normal state of any Unix process which currently does
nothing (e.g. is waiting for input from somewhere). I'm not sure what
Etime and Stime means - I don't know Solaris. The Time column tells you
how much CPU time this process has used since it was created.
In PostgreSQL you have one postgres process for each connection. So you
can see when the connection was opened by looking at the process'
creation time.
HTH,
Tino.
--
www.craniosacralzentrum.de
www.spiritualdesign-chemnitz.de
Tino Schwarze * Lortzingstraße 21 * 09119 Chemnitz