RE: Weird database hanging
- From: Amar Kumar Padhi <amar.padhi@xxxxxxxxx>
- To: oracle-l <oracle-l@xxxxxxxxxxxxx>
- Date: Tue, 18 Sep 2007 02:30:37 +0400 (GST)
Is the web app using Connection polling? No thing bad about this, but rules can
be defined to open up a set of connections in advance, this could result in
the spike.
thanks!
amar kumar padhi
www.amar-padhi.com
pushed from PDA
-----Original Message-----
From: "Sunil Kanderi" <sunil.kanderi@xxxxxxxxx>
To: "oracle-l" <oracle-l@xxxxxxxxxxxxx>
Sent: 18/09/07 01:46
Subject: Re: Weird database hanging
I have gone through all the responses to this problem and this seems
very similar to a problem that I am seeing on one of my 9i database. We
have lots of web applications running from that database and once in a
while there is a sudden spike in the number of connections from a
Oracle user related to one of the web applications and the concurrent
connections for that particular web application user jumps from 10 or 15
concurrent connections to 100s in less than a couple of minutes. The
database and the server hang until these connections are cleared. At
times we cannot even get a connection to the server when this happens. I
usually don't see any alerts in the alert log when this happens.
Thanks,
Sunil.
Don Seiler wrote:
> Running 10gR2 (10.2.0.2) on RHEL4 on 4 64-bit dual-core box. Earlier
> today users complained en masse about not being able to connect to the
> database. I got to a term and found that even trying to log in
> locally as sysdba was hanging.
>
> The alert log had this:
> Fri Sep 14 11:23:36 2007
> kkjcre1p: unable to spawn jobq slave process
> Fri Sep 14 11:23:36 2007
> Errors in file /u00/app/oracle/admin/foo/bdump/foo_cjq0_11519.trc:
> Fri Sep 14 11:25:46 2007
> kkjcre1p: unable to spawn jobq slave process
> Fri Sep 14 11:25:46 2007
> Errors in file /u00/app/oracle/admin/foo/bdump/foo_cjq0_11519.trc:
> Fri Sep 14 11:26:49 2007
> PMON failed to acquire latch, see PMON dump
>
> I hit the Google but before I came back, my login was ready and things
> seemed back to normal. Checking back in the alert log I saw a lot of
> these:
>
> WARNING: inbound connection timed out (ORA-3136)
>
> I'm assuming those are just symptoms of the original cause, and I've
> already found suggestions for treating timeouts in sqlnet.ora and
> listener.ora.
>
> The trace files mentioned have lines such as:
>
> Waited for process J000 to initialize for 60 seconds
>
> repeated for 70, 80, 90 seconds, etc.
>
> So in summary, there was hanging for everyone, it seemed to clear
> itself up. One post on Google suggested that the PROCESSES init
> parameter might be too low. Any other possibilities?
>
>
--
//www.freelists.org/webpage/oracle-l
--
//www.freelists.org/webpage/oracle-l
Other related posts:
