Cinder volume fails to transition to reserved state after instance shelving due to stale attachment
Issue
- After shelving an instance, the associated Cinder volume does not move into the
reservedstate. - The cinder volume_attachment table contains two overlapping records for the same volume:
- One with status
attached(stale/old) - One with status
reserved
- One with status
- The affected volume is a boot volume, therefore:
- The volume attachment cannot be deleted directly with the Cinder CLI.
- The volume cannot be detached from the instance through the Nova API.
Environment
- Red Hat OpenStack Platform 17.1
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.