Re: Is there any way to push Standby database to delete archivelogs quicker (ARCHIVELOG DELETION POLICY in use)?

Hello everyone,

Just to give you a better idea why I am looking for something like that:
- A customer uses a file system to store archivelogs on Standby side
- In order to let Oracle to manage those automatically we decided to
use FRA and RMAN ARCHIVELOG DELETION POLICY. This would allow to avoid
any custom scripts implemented on standby side   for that purpose.
This works very good with one exception.

The customer is a big shop with huge commonly used infrastructure
around different systems. There are a standard monitoring procedure
for all File Systems out there. If FS is 95% (X%) full the monitoring
system starts to send alerts. If we use FRA Oracle tends to use 100%
of space and delete applied archivelogs if running our of space only.
The proper way of handling this issue probably would be introducing a
new FS check routine which would use V$RECOVERY_FILE_DEST;
V$FLASH_RECOVERY_AREA_USAGE views for monitoring purposes. The problem
is that as in any BIG organizations nobody willing change existing
procedures and if even would like to change it would take a
significant amount of time (BTW: Oracle is just 10% of RDBMS systems
used in that organisation).
If we say this isn't an option in that case what would be possible solutions?

I see 2:
- Find the way to let Oracle to clear FRA earlier then FS check
threshold is reached. (This is what I asked in first post).
- I don't like it: Lower db_recovery_file_dest_size parameter to be
95% (x%) of a file system size. Oracle DBA still will have some space
to increase parameter back and work on the problem when system will
hit 100% utilization of FRA.  The obvious disadvantages are: the
standard monitoring will not alarm when FRA running out of space,
primary DB might stop until the FRA is cleaned.

What is your experience in monitoring FRA utilization (excluding
V$RECOVERY_FILE_DEST; V$FLASH_RECOVERY_AREA_USAGE option) if there are
any?

Best regards,
Yury


On Wed, May 20, 2009 at 6:22 PM, Jurijs Velikanovs
<j.velikanovs@xxxxxxxxx> wrote:
> Hello dear List members,
>
> Just wonder if someone came across the following question:
> 10.2.0.4 Data Guard configuration
> Backups are taken on primary. As soon as an archivelog applied on a
> standby it could be deleted.
> ON STANDBY:
> log_archive_dest_1='LOCATION=use_db_recovery_file_dest'
> CONFIGURE ARCHIVELOG DELETION POLICY TO APPLIED ON STANDBY;
>
> Oracle deletes applied archivelogs in a "lazy" mode.  I assume as soon
> as space usage reach some threshold of db_recovery_file_dest_size
> Oracle starts to delete some "old"/applied archivelogs from fra.
> There are the following messages in the alert.log file:
> Deleted Oracle managed file
> +DATA/gtst/archivelog/2009_06_16/thread_1_seq_83.349.689686855
> Deleted Oracle managed file
> +DATA/gtst/archivelog/2009_06_16/thread_1_seq_84.351.689687247
> ...
>
> Just wonder if someone knows any way to change the "threshold" to push
> Oracle to delete applied archivelogs faster? Let say do not keep those
> at all on keep a small amount of those?
>
> Thank you in advance,
> Yury
>



-- 
Jurijs
+371 29268222 (+2 GMT)
============================================
http://otn.oracle.com/ocm/jvelikanovs.html
--
http://www.freelists.org/webpage/oracle-l

Other related posts: