To upgrade cluster nodes with High Availability configurations, always make sure that the Cluster nodes are in sync or Cluster sync status is healthy.
Steps to Upgrade Cluster nodes with High Availability configurations:
1. Always Upgrade Publisher First, followed by upgrading of remaining cluster nodes/subscribers.
2. It will prompt to turn off the Publisher Standby Configuration before performing upgrade. This is
done so that, Designated Standby should not get promoted as new publisher.
3. Once the Designated Standby configuration is set to False, Upgrade of Publisher succeeds.
4. Now try upgrading remaining cluster nodes.
Suppose you unknowingly upgraded a cluster of nodes, where designated standby is out of sync, the
cluster sync status may end up something like below:
Problem:
On upgrading an out of sync cluster with VIP and Designated Standby enabled, following is the state:
On Publisher: Cluster Sync status for Subscriber Node (Designated Standby) = IN_PROGRESS
On Subscriber: Cluster Sync status = Out of Sync.
This means, subscriber cannot be joined back to cluster, until the same is dropped from Publisher.
Solution:
Drop subscriber from Publisher.
1. On Publisher, Drop subscriber will be refused, and an error is prompted to disable Designated
Standby configuration.
2. Next set Designated Standby configuration to False using Cluster Wide parameters.
3. If subscriber node is part of VIP configuration, then Drop subscriber will be refused again, and an
error is prompted to disable VIP configuration.
4. After disabling the above, Drop subscriber completes successfully on Publisher.
5. Since Cluster is out of sync, manually reset the database on Subscriber and perform Make
subscriber on this node.
6. Once the subscriber is added back to cluster, reconfigure the VIP and Designated standby
configuration as before to have high availability working as expected.