We are upgrading Longhorn from 1.1.2 to 1.2.2. After the upgrade of the system we saw that some instance managers were still running the old version. It looks like this is expected:
Q1. What is the best way to move all pods to the new instance managers? None of the pods using persistent volumes are critical so it’s ok to have a bit of a disruption for them. What we did to force it was to cordon / uncordon each node which seemed to work but is disruptive for the rest of the pods that aren’t using persistent volumes.
After switching over to the new instance managers we saw that the previous engine-image-ei Longhorn pods were still running. It looks like this is supposed to be automatically cleaned up after 60m:
Q2. What is the best way to clean this up? Can the daemon set just be deleted or is something else needed?