I'm wondering about best practices and recommendations for managing Oracle logs on 10g Real Application Cluster cluster nodes in Linux. Using 'logrotate' is likely the best method, but there are many options and it would be nice to know of any recipes that use logrotate, and/or other methods of archiving.
In general, I'd like to keep logs available for 30 days and send them to tape afterward, but there are many types of log files on a cluster node, and I'd like to know the recommendation for keeping and archiving each.
I've looked in several books, and the best information I've seen was in the RAC Handbook, which explains the log file directory structures, but nothing on managing them. Also that the alert log is opened and closed with each write, but other logs may remain open as long as a process is alive, which might cause problems for archiving.
ALL the Oracle homes
You need to review ALL the Oracle homes and periodically truncate the logs in CRS, ASM and RDBMS homes.
For logs that remain open as long as a process is alive, use the copytruncate option from logrotate.
"The person who says it cannot be done should not interrupt the person doing it." --Chinese Proverb
Click Here to Expand Forum to Full Width