We are planning a refresh of our Discovery MID Servers. Our schedules are organized by location, and our main concern is around credential and IP service affinities. If these are lost, each new MID Server may attempt multiple ports and credentials until affinities rebuild, creating a large spike in login attempts.
We’ve considered three possible approaches:
- Slowly introducing new MIDs by replacing one MID per cluster each week or so, giving affinities time to rebuild gradually.
- Creating temporary schedules by device type, using a single credential alias to ensure only the correct credential is attempted, allowing affinities to rebuild in a controlled way. Once complete, we would remove the temporary schedules. (Obviously, this would be a tedious process)
- Updating sys_ids in the affinity tables to point to the new MID Servers.
Our questions:
What are the best practices for refreshing MID Servers and managing affinities - should old entries be purged, and allowed to rebuild naturally?
Are there recommended maintenance practices for the affinity table?
Ultimately, what is the best way to refresh MID Servers without causing unnecessary login attempts or disruption? Or are we overthinking the impact?
