Volume drivers should support check is volume already exists

Registered by Ivan Kolodyazhny

After some fail or cinder-volume restart we want try create/delete/resize volume again. To make it more reliable volume drivers should to check if backend already created a volume (E.g. check_volume_exists method)

Blueprint information

Status:
Complete
Approver:
None
Priority:
Undefined
Drafter:
Ivan Kolodyazhny
Direction:
Needs approval
Assignee:
Yuriy Nesenenko
Definition:
Obsolete
Series goal:
None
Implementation:
Unknown
Milestone target:
None
Completed by
Sean McGinnis

Related branches

Sprints

Whiteboard

Sean: Abandoning this as it appears to have stalled in reviews. If still desired we can reopen.

Avishay: I would change the semantics of the call to be "what is the status of volume X"? The return value would be one of: OK, not found, error (I think these are states everyone can agree on). Now you can have checks to make sure the Cinder DB is consistent with the storage. This will reduce cases where the volume exists in one but not the other, but also give some indication of the volume's health. Even better, make a batch call to the driver with a list of volume ID - "what is the status of these 100 volumes?"

Gerrit topic: https://review.openstack.org/#q,topic:bp/check-volume-exists,n,z

Addressed by: https://review.openstack.org/174823
    Checking the existence of volume

Addressed by: https://review.openstack.org/175809
    Implement blueprint: check-volume-exists

(?)

Work Items

This blueprint contains Public information 
Everyone can see this information.

Subscribers

No subscribers.