Clusterware maintain Voting disk to know heartbeat of the nodes in a cluster.
When ever node joins a cluster, voting file updated. When ever voting file is updated, Clusterware software starts communicating with node.i.e CRS demon on one node will communicate with CRS demon on another node. Then On each node, clusterware software starts its services and resources.
Technically CSSD monitor will update Voting files on each disk inside the diskgroup. For every second CSSD process updates (Node availability information) to the voting files. Atleast each node has to access 2/3 voting files. If any node is unable to access 2/3 of voting files, it evicts from cluster. OPROCD kills the node in a cluster to protect data integrity. From 11GR2, OCR will backup the votingfiles. If votingfiles are corrupted, we can recover using backup taken by OCR.
We know that Votingfiles are stored on ASM, we can find Out using Following Query
[root@tnc1 ~]# crsctl query css votedisk
## STATE File Universal Id File Name Disk group
-- ----- ----------------- --------- ---------
1. ONLINE 4bbee2aa15204fb5bf311cc8ebcedd10 (/dev/sdh2) [NEVOT]
Located 1 voting disk(s).
Currently we have only 1 Voting file on the cluster. If this Voting file is corrupted, All nodes in a cluster will evict from the cluster.
To Stabilize RAC Environment, Multiplex Voting Files in Oracle RAC.
To Multiplex Voting files, Create a new diskgroup using "ASMCA or SYSAM" with NORMAL Redundancy or High redundancy and Replace Voting files from old diskgroup to New diskgroup;
OLDDISKGROUP: NEVOT
NEWDISKGROUP: GRID
Make sure, New disk group is mounted on all instances
SQL> select name,group_number,state from v$asm_diskgroup;
---
Replace Voting files from One OLD ASM diskgroup (External Redundancy) to another NEW ASM diskgroup (High Redundancy). perform operation as root user from any node of the cluster
Successfully, Voting files are Replaced / Multiplexed on Oracle RAC Cluster
--Nikhil Tatineni--
--12c Oracle RAC--