Recommended procedure for vRealize Automaton (vRA) service or node restart in a 3 node cluster
Usually, when dealing with a 3-node cluster, it doesn’t matter which node is the master one. In such cases reboot of the appliances should happen always one by one with making sure each appliance has fully loaded before initiating the next reboot. One should make sure there are always 2 (two) running appliances. This ensures quorum is met and environment is fully operational.
There are three main scenarios when doing a reboot of vRA:
- Reboot only the master appliance
- Make the replication Async.
- Initiate reboot on the master node.
- Wait for the appliance to fully load up.
- Make the replication Sync again.
- Reboot any of the secondary appliances
- Reboot the passive appliance.
- The potential appliance now becomes passive.
- Wait for the rebooted appliance to fully load up.
- The once passive appliance now becomes a potential one.
- Reboot the currently passive one.
- Once it’s rebooted it will be automatically switched to the potential role and all roles should remain the same
- Reboot all three appliances
- Turn the replication to Async mode.
- Reboot the master node.
- Wait about 30 to 60 seconds and then reboot the two other nodes.
- After all servers are fully loaded (all services appear as “Registered”) turn the replication to Sync mode again.