Error migrating volumes inside same storage center
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Cinder |
Invalid
|
Undecided
|
Unassigned |
Bug Description
I have the following situation:
There are two cinder-volume services on different servers using the same storage backend (a Dell SCv2020)
Whenever I try to migrate one volume to another server I get the following errors:
1. The new volume is created and attached to the instance but it fails to remove the old volume, despite appearing with different IDs the logs says it can't delete since there are two volumes with same ID and the old volume get stuck on the faled to delete status.
2. Since the old vaolume is not useful anymore, I change it status to available, it can't be deleted because of the problem above, but when I unmanage the problematic volume, the storage manage client shows that the attached volume is marked as unmanaged.
To add a bit of context, I had the cinder-volume on every compute server and decided to put the service on the controller node and remove from compute nodes.
When you say "migrate one volume to another server", do you mean one cinder-volume server to the other cinder-volume server? That could definitely cause issues if you have both cinder-volume services managing the same SC backend.