Re: 18XE memory leak in CJQ0
Date: Mon, 25 Nov 2019 09:46:13 +1100
Message-Id: <DC0ECA67-2044-4BE5-BD5F-D70FF92994B4_at_gmail.com>
Hi Rich,
Check out bug 28831971.8
Cheers,
Leng
> On 23 Nov 2019, at 1:51 am, Rich J <rjoralist3_at_society.servebeer.com> wrote:
>
>
> Hey all,
>
> I had noticed that RAM usage was slowly increasing on the OL7.4 server running 18 XE (and a host of other things) and tracked it down to the CJQ0 process. Recording PGA usage hourly as shown in V$PROCESS for the past 40+ days, the PGA_ALLOC_MEM of the CJQ0 process increased by 4.2MB/day on average with little deviation. That sure smells like a memory leak to me. The XE instance has the sole user PDB, with only the default jobs in the CDB and APEX and 1 other job scheduled in the PDB. JOB_QUEUE_PROCESSES was the default of 4000 (seems a tad high for a default).
>
> I reduced the JOB_QUEUE_PROCESSES to 4 and killed the "xe_cjq0_XE" process from the OS to give it a fresh start. Related MOS articles had mentioned the killing as a workaround, which seems like it would be one permanent solution for an XE installation.
>
> Anyone run into this in EE, either 18 or 19? I'm looking to move to 19 soon with an instance that will have more scheduled jobs and a much larger SGA that may mask this issue.
>
> Thanks,
> Rich
-- http://www.freelists.org/webpage/oracle-lReceived on Sun Nov 24 2019 - 23:46:13 CET