Hi all, in a Oracle RAC 11g i see a lot of audit file per second.
The cause is the CHECK_INTERVAL parameter sets on the ora.name.db resource of the cluster that is equal to 1 second.
The only workaround is to increase its value or proceed to clean the filesystem periodically.
My question is: there is someone of you that have a cluster with this problem?
Is safe increase the time of checking maybe around 30 second?
Is there any other solution?
Is not wrong the CRS checking every second, is wrong the logging way; one aud file for every check is too much.
I know we talk about two different systems; database and cluster service and when a database is accessed by a SYSDBA account well... it log!
So every check made up by cluster is done with SYSDBA user and for this reason i say "it's wrong".
How can i know when i've a SYSDBA connection different from the "cluster connection"?
The adump directory become useless...
I won't be polemical, i'm only speaking...
Agreed. By default all connections with SYSDBA/SYSOPR are audited and there is no way to disable it.
As of now there is no separate user for CRS to check Database health and all connection are as sysdba only. Ideal way can be for monitoring it can be without sysdba and for start/stop it can be SYSDBA . This functionality is not present as of now and I am not sure if there is a way to fix it. May be to check with Oracle Support if they have plan to introduce this feature
AUDIT_SYS_OPERATIONS Set To FALSE Yet Audit Files Are Generated (Doc ID 308066.1)