Test System Upgrade with New Instance Manager
- Prepare 3 sets of longhorn-manager and longhorn-instance-manager images.
- Deploy Longhorn with the 1st set of images.
- Set
Guaranteed Instance Manager CPU
to 40, respectively.
Then wait for the instance manager recreation.
- Create and attach a volume to a node (node1).
- Upgrade the Longhorn system with the 2nd set of images.
Verify the CPU requests in the pods of both instance managers match the settings.
- Create and attach one more volume to node1.
- Upgrade the Longhorn system with the 3rd set of images.
- Verify the pods of the 3rd instance manager cannot be launched on node1 since there is no available CPU for the allocation.
- Detach the volume in the 1st instance manager pod.
Verify the related instance manager pods will be cleaned up and the new instance manager pod can be launched on node1.
[Edit]