Prev: Agent 10.2.0.2 patchset install on AIX
Next: "reliable message" wait event
From: sealo on 4 Jan 2007 21:30
Hello,
Now the ORACLE was often down every one week. But this issue had not
happened before. And checking the system monitor log, it shows that the
memory was used up.

And, according to the alert log, it shows that the oracle was dead
because
* kkjcre1p: unable to spawn jobq slave process *
Except this, there are no abnormal log in the log.

And I check the current running status, the process ora_j000 have
already used 240M memory, and the usgae was still increasing.

I also check the DBA_Jobs, all the jobs are in normal status, no break.

I guess, the memory was used up by the ora_j000. Could you give me
something method to find the root cause?


Environment

ORACLE Release 10.1.0.2.0
Linux: 2.4.20

From: Mladen Gogala on 5 Jan 2007 00:33
On Thu, 04 Jan 2007 18:30:02 -0800, sealo wrote:

> Hello,
> Now the ORACLE was often down every one week. But this issue had not
> happened before. And checking the system monitor log, it shows that the
> memory was used up.
>
> And, according to the alert log, it shows that the oracle was dead
> because
> * kkjcre1p: unable to spawn jobq slave process *
> Except this, there are no abnormal log in the log.
>
> And I check the current running status, the process ora_j000 have
> already used 240M memory, and the usgae was still increasing.
>
> I also check the DBA_Jobs, all the jobs are in normal status, no break.
>
> I guess, the memory was used up by the ora_j000. Could you give me
> something method to find the root cause?


You have an old version of RH EL 3.0 and you have an ancient
version of Oracle. You should update AQAP, but you probably know
that already. My advice is to upgrade to RH EL 4.0 and Oracle 10.2.0.3

Besides V$PROCESS_MEMORY, V$PGASTAT and V$SESSTAT, I'd also use pmap -x



--
http://www.mladen-gogala.com
 | 
Pages: 1
Prev: Agent 10.2.0.2 patchset install on AIX
Next: "reliable message" wait event