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

Review your Oracle & HP MC/ServiceGuard configuration

Review your configuration of Oracle and MC/ServiceGuard to avoid headaches.

A common problem that I see at clients is with the configuration of HP MC/ServiceGuard and Oracle. With some minor changes to the configuration of Oracle and MC/ServiceGuard, you can prevent a lot of headaches.

One of the biggest headaches is the configuration of the monitoring of Oracle that triggers a package failover. A common mistake is triggering package failover when a database is down. One of the sample scripts for MC/ServiceGuard and Oracle does just that. I'm not really sure what kind of database situation that this will rectify.

In my experience, the database usually down because the DBA shut it down (for system maintenance, parameter changes, etc.) or there is a serious problem with Oracle. Failing the database over to another UNIX machine is not going to correct 99.9% of the problems with Oracle that cause the database to shutdown. In most cases, failing the package over to another machine is just going to add 15 minutes to the downtime, while the DBA waits for the package to start up on a failover node in the ServiceGuard cluster. Your best bet is disable package switching based on an Oracle database going down, and rely on the heartbeat connection to initiate package failover.

Another area to be careful with is with the NET8 listener. Make sure the you uniquely name your listeners. To accomplish this, change the section that begins with LISTENER to LISTENER_FIN, and the SID_LIST_LISTENER section to SID_LIST_LISTENER_FIN. I like to keep the LISTENER part of the name, and suffix it with a word(or part of a word) that identifies the purpose of the databases that this listener is serving. I like to use a centrally located listener.ora file (/etc/listener.ora), therefore, uniquely identified listeners is the only way to go.

The final part of the MC/ServiceGuard configuration that you should be aware of is the oratab file. Make sure that all databases from all machines in the cluster are defined in your /etc/oratab file. Also make sure the you have set flag in the third column for each database to 'N', so that dbstart, and dbshut won't try to startup or shutdown these databases, the MC/ServiceGuard scripts should be handling that.

For More Information

  • What do you think about this tip? E-mail us at [email protected] with your feedback.
  • The Best Oracle Web Links
  • Have an Oracle tip to offer your fellow DBA's and developers? The best tips submitted will receive a cool prize--submit your tip today!
  • Do you have any technical questions about Oracle administration or development? Post them--or help out your peers by answering them--in our live discussion forums.
  • Check out our Ask the Experts feature! Our Oracle gurus are waiting to answer your toughest Oracle questions.

Dig Deeper on Oracle database design and architecture