Cinder does not start after volume node reboot
Bug #1044312 reported by
Unmesh Gurjar
This bug affects 2 people
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Cinder |
Invalid
|
Undecided
|
Unassigned |
Bug Description
Branch: master (commit ID: cabb54c5806022e
Scenario:
1. On a running devstack installation (having Cinder Volume service), stop the services using "unstack.sh" script.
2. Reboot the node.
3. After the node comes up, start all the services (including Cinder).
Actual Result:
Cinder volume service fails to start with the error volume group "stack-volumes" does not exist.
The volume group "stack-volumes" gets deleted ('vgdisplay' does not list it). Hence, all the volumes were lost.
Expected Result:
Cinder Volume should start up successfully (and the volume group should remain intact).
Stack Trace: http://
To post a comment you must log in.
Turns out this is actually not related to unstack or cinder per-say. The volume group is removed on the reboot operation, running unstack.sh it will remove the targets that have been created, however you can do a vgdisplay and see that stack-volumes is still there.
Rebooting however the system will come back up and stack-volumes will be removed.