Problem solve Get help with specific problems with your technologies, process and projects.

Two alert.log files not in sync

Here's some background: the configuration I will refer to consists of two Sun V880s in an active-passive failover cluster on Veritas Quickstart and Veritas FileSystem. The machines have redundant D2 disc arrays that are striped and mirrored. One machine is active and takes care of all transactions, and this is the machine that hosts the Oracle 8i ( database. In the event of a failure, the Veritas Quickstart shuts the Oracle instance down via the scripts, dismounts the shared disk array and re-mounts the array on the secondary machine prior to firing the Oracle up. Oracle is therefore only active on a single machine at any given point in time. The other machine is dormant and awaits a failure or manual switch-over.

There happens to be 2 (!!) alert_<oracle_sid>.log files, and they seem to contain contradictory information. The first file, and one I typically check, resides in the /opt/oracle/admin/<oracle_sid>/bdump directory. This file contains the varied information from the Oracle subsystem that includes shutdowns/startups/logswitches/extend changes etc. The second alert_<oracle_sid>.log file resides in the /opt/oracle/product/ directory. This latter file only contains multiple entries that record the shutdown (abort) of the opposite fail over partner machine.

That is, when we perform a manual switch with the "vcsqs -switch" command, the alert.log in the first directory, on the initially active machine, indicates a normal shutdown, while the second alert.log of the same machine has no entry of this event. If one, however, connects to the new active machine, after the switch completes, and check the first alert.log file, it shows a startup event. Should you check the second alert.log file, you find that it indicates a shutdown (abort) event at the approximate time the failover was initiated. Why is this? The Oracle on the initially dormant machine was NOT active, so why the shutdown abort event in the log? I have asked Veritas whether their software does not perhaps write an entry to that second logfile, but they have assured me that Veritas NEVER writes any entries to the Oracle logs. It looks really bad if the customer looks at those logs and it appears as if we are doing aborts to switch over, but this is not the case at all. This has me mystified, and I would sincerely like to know why this occurs and what we can do to appease the worried customer.

I'm not exactly sure why you are exhibiting this behavior. But maybe some information can help you.

The alert log in /opt/oracle/{sid}/bdump is only written here because the instance has read your parameter file and noted the value you set for the BACKGROUND_DUMP_DEST parameter. If this parameter is not set, or if the instance could not find the parameter file, then by default, the alert log is recorded in $ORACLE_HOME/rdbms/log. So it doesn't surprise me that Oracle is writing to two different locations, at least at two different times.

For More Information

Dig Deeper on Oracle database design and architecture

Have a question for an expert?

Please add a title for your question

Get answers from a TechTarget expert on whatever's puzzling you.

You will be able to add details on the next page.

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.