This content has been marked as final. Show 4 replies
Please check instance alert log files for error messages why instance is going down.
There are multiple causes for these issues and exact messages will help to further narrow down the issue. You can also download raccheck utility from MOS to make sure all best practices are being followed.
As you are facing node eviction issue, to find out reason please share following details:
1. clusterware alert log file
2. cssd.log file
3. crsd.log file
4. No. of voting disks (*sometimes if you have only one voting disk, one node will be evicted automatically as i have observed*)
is it the same node all the time?
I have seen nmccollector (OEM using Memory Collection vs SQL Collection) chew up swap and causing the instance to die/restart.
use "top" to see if you have sufficient swap space.
Database Alert logfile will be first point of diagnosis. Check database alert log files when instance get restarted and upload their contents.