During a rolling upgrade, volume creation can fail with 'volume_type cannot be None'
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Cinder |
Fix Released
|
Undecided
|
Michal Dulko |
Bug Description
During the rolling upgrade process, the first things to be upgraded (I believe) are the database and the c-api. After these two steps it's assumed that the cloud should still be functional, given the rpc version pinning and OVO work.
At this point in the upgrade it's possible for volume creation to fail, saying that volume_type cannot be None. This is because in mitaka volume creation with a type of None was supported, but in master a commit changed the way this worked, so https:/
Perhaps some check should be added in the API side to ensure that an exception won't be thrown if the scheduler is still in mitaka, or the fix linked above could be backported to mitaka to mitigate this.
Changed in cinder: | |
assignee: | nobody → Nate Potter (ntpttr) |
Fix proposed to branch: stable/mitaka /review. openstack. org/363928
Review: https:/