Ubuntu Cloud Archive

  1. Series bobcat
  2. Bug #2019190
  3. Comment #13

Comment 13 for bug 2019190

Revision history for this message
Nishant Dash (dash3) wrote : Re: [RBD] Retyping of in-use boot volumes renders instances unusable (possible data corruption)

Hello,

I am hitting this issue on a production cluster (jammy-yoga).

At the moment I have 2 situations with workarounds other than cold migrating:
- Instances that been retyped but not rebooted so are active.
For these, I tried to shelve and unshelve the VM and that was enough to update the block-device-mapping entry in the nova-db for the instance. (Cold migration works too like described in comment #11)

- Instances that have been retyped and sent a reboot request and are in ERROR.
For these, setting their state to ACTIVE and then doing the shelve-unshelve works.

These are the only workarounds I have so far.