Re: auto shutdown of archiver on startup

From: Charles Hooper <hooperc2000_at_yahoo.com>
Date: 9 Feb 2007 14:00:01 -0800
Message-ID: <1171058401.141297.176190@a75g2000cwd.googlegroups.com>

On Feb 9, 3:57 pm, "EdStevens" <quetico_..._at_yahoo.com> wrote:
> Platform : Oracle 10.2.0 on HP-UX
>
> Just created a new, very vanilla database with scripts genned out of
> dbca (custom database). Maybe this was happening on previous
> databases and I just never noticed, but I see that several seconds
> after the database opens, we see this in the alert log:
>
> Fri Feb 9 12:02:31 2007
> Completed: ALTER DATABASE OPEN
> Fri Feb 9 12:02:32 2007
> db_recovery_file_dest_size of 16384 MB .... <snip>
> Fri Feb 9 12:03:29 2007
> Shutting down archive processes
> Fri Feb 9 12:03:34 2007
> ARCH shutting down
> ARC2: Archival stopped
>
> Nothing else appeared amiss, so I monitored the processes during a
> startup sequence:
>
> stevense> ps -ef|grep ora _arc|grep -v grep
> oracle 24278 1 0 12:23:55 ? 0:00 ora_arc1_<sid>
> oracle 24276 1 0 12:23:55 ? 0:00 ora_arc0_<sid>
> oracle 24280 1 0 12:23:55 ? 0:00 ora_arc2_<sid>
> stevense> ps -ef|grep ora _arc|grep -v grep
> <snip several reps of same, then>
> stevense> ps -ef|grep ora _arc|grep -v grep
> stevense>
> oracle 24278 1 0 12:23:55 ? 0:00 ora_arc1_<sid>
> oracle 24276 1 0 12:23:55 ? 0:00 ora_arc0_<sid>
> stevense> exit
>
> So, it looks like three archivers are getting started at db startup,
> then one is almost immediatey shut down at the end of the startup
> cycle. I have log_archive_max_processes=2. I know that "LGWR process
> starts a new ARCn process whenever the current number of ARCn
> processes is insufficient to handle the workload." But I haven't seen
> definitive docs to say it will also decrease the number of ARCn
> processes when workload lightens, and I am somewhat surprised that
> there would be need to be doing this dynamic re-allocation right at
> startup.

I see something similar here on 10.2.0.2 (prior to monthly security patches)

>From the Alert Log of one of my databases:

Tue Jul 11 20:09:04 2006
Completed: alter database open
Tue Jul 11 20:09:06 2006
Thread 1 advanced to log sequence 2155
Current log# 1 seq# 2155 mem# 0: ....LOG Tue Jul 11 20:09:07 2006
Shutting down archive processes
Tue Jul 11 20:09:09 2006
db_recovery_file_dest_size of 60000 MB is 58.43% used. Tue Jul 11 20:09:12 2006
ARCH shutting down
ARC2: Archival stopped

>From the sid_arc2 trace file from the same time period:

2006-07-11 20:09:05.593
SERVICE NAME:(SYS$BACKGROUND) 2006-07-11 20:09:05.546
SESSION ID:(220.1) 2006-07-11 20:09:05.546
2006-07-11 20:09:05.546 2525 kcrf.c tkcrf_clear_srl: Started clearing Standby Redo Logs
2006-07-11 20:09:05.812 2833 kcrf.c tkcrf_clear_srl: Completed clearing Standby Redo Logs

My best guess is that Oracle created an archiver process to clear the Standby Redo Log, and then terminated the archiver process after the Standby Redo Log was archived. I have not restarted the Oracle instance since applying the October 2006 patch, so I cannot confirm if this logging detail is still present after the patch.

Charles Hooper
PC Support Specialist
K&M Machine-Fabricating, Inc. Received on Fri Feb 09 2007 - 16:00:01 CST