If you suspect that your Clusterware is not working properly and you have these conditions:
- Disk mapping is the same.
- Disk major and minor are the same.
- Instead of one cluster you have two different clusters.
- You can't add a new node on the cluster.
Then you have to verify if your storage is shared.
This is the procedure. Please execute it on all the nodes:
1. Stop the Clusterware:
crsctl stop crs -f
2. Record the name of the vfstart and vfend of the Voting Disk:
kfed read /dev/oracleasm/disks/VOTE2|grep vf kfdhdb.vfstart: 320 ; 0x0ec: 0x00000140 kfdhdb.vfend: 352 ; 0x0f0: 0x00000160
3. Perform a dd dump according to the above numbers:
dd if=/dev/oracleasm/disks/VOTE2 bs=1048576 skip=320 count=33 |od -c > /tmp/dd<node>.log
4. Gather the results from previous steps and compare the logs:
diff /tmp/dd<node1>.log /tmp/dd<node2>.log
If there is any difference between the logs, then the Voting Disk is not shared between the nodes. In that case you have to contact the system administrator to check further.