RE: Archived logs in wrong destination with 2-node RAC

From: Harel Safra <harel.safra_at_gmail.com>
Date: Wed, 28 Jul 2010 20:41:37 +0300
Message-ID: <03b701cb2e7c$23dbc960$6b935c20$_at_gmail.com>

When you shut down the instance you don't disable the redo thread it's writing.

One of the other instances will archive redo logs from that thread in some situations, I have seen it happen when the active instance has wrapped around all its' redo logs.  

For this reason and faster recovery considerations I like to use a central, shared, location as an archive log destination for all the instances. In our case it's a NetApp volume in NFS.  

Harel Safra  

From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of daniel.hubler_at_aurora.org Sent: Wednesday, July 28, 2010 8:23 PM
To: oracle-l_at_freelists.org
Subject: Archived logs in wrong destination with 2-node RAC  

Oracle 10.2.0.3
AIX 5.3.8.3 + HACMP 5.4.1
2-node RAC

Archived log file systems are defined to the local node only; but failover via HACMP.

When we shutdown the database instance on node-2, (not CRS....just the DB),
an archive log file is generated from that instance (node-2), but shows up in the archive log destination on node-1.

Happens every time.
Does not happen to archived logs that are generated via normal operations, or even a "alter system switch logfile".

Not a new installation.
Is a test environment where we turned on archive logging not too long ago.

Not certain why that archive log is showing up in the other/incorrect side.

Any ideas?

--
http://www.freelists.org/webpage/oracle-l

Received on Wed Jul 28 2010 - 12:41:37 CDT