Exadata: DB Node Evictions when stopping Clusterware on multiple nodes at the same time.
Last updated on JULY 24, 2013
Applies to:Oracle Database - Enterprise Edition - Version 126.96.36.199 to 188.8.131.52 [Release 11.2]
Information in this document applies to any platform.
Shutting down the clusterware on all the nodes of the cluster at the same time using "crsctl stop cluster -all" or "crsctl stop crs" with dcli command, would evict multiple nodes from the cluster. This happens only on clusters with more than 2 nodes and there are application VIPs configured VIPs in the private/IB network.
Non exadata customer's can also run into the same issue, if they have configured application VIPs for the cluster interconnect.
Looking at the CSSD logs, we will see that the nodes lost connectivity with each other and getting evicted as part of the split-brain resolution. And the network error is reported for the IP address, which is defined as the application VIP.
Added application VIPs to the cluster private network
Sign In with your My Oracle Support account
Don't have a My Oracle Support account? Click to get started
My Oracle Support provides customers with access to over a
Million Knowledge Articles and hundreds of Community platforms