yahoo-eng-team team mailing list archive

[Fei Long Wang <flwang@xxxxxxxxxxxxxxx>]

** Changed in: nova
       Status: Confirmed => Fix Released

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to OpenStack Compute (nova).
https://bugs.launchpad.net/bugs/1367968

Title:
  Data loss on reboot following Ceph volume-backed instance rebuild

Status in OpenStack Compute (Nova):
  Fix Released

Bug description:
  Under particular circumstances, volume backed instances that are
  rebuilt may revert to a pre-rebuild state, losing any data added since
  that rebuild.

  Our environment:

  * OpenStack Havana
  * Ceph Emperor Backend
  * Using Ceph for Glance, Cinder and Object Store

  All instances were created and started through Horizon.

  Steps to reproduce:

  * Create a new instance, using Instance Boot Source: Boot from image (creates a new volume).
    - Note at this point the instance details will show Image Name: (not found)
  * Log in through SSH, create some identifiable pre-rebuild test files.
  * Rebuild Instance, selecting the same original image.
    - Note at this point the instance details will show, for example, Image Name: ubuntu-12.04-x86_64
  * Log back in through SSH. As expected the test files should not be present.
  * Create a second set of post-rebuild test files.
  * Shut down the instance from inside itself - "shutdown -h now".
    - relevant note - bug was *not* triggered by a soft restart.
  * Start the instance again from Horizon.
  * Log in. The post-rebuild test files will no longer be present, the pre-rebuild test re-appear. Has reverted to pre-rebuild state.

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1367968/+subscriptions