Open Stack

Folks,

I have question, I am seeing a behavior, while deleting nova instance it returns status as an Error and fails to delete itself.
So scenario, where I see this behavior is as follow:

  1.  Create a nova VM instance.
  2.  Create a cinder volume
  3.  Attach this volume to nova vm instance and after that wait for volume to go to ‘In-use’ state.
  4.  Mount the created partition to VM
  5.  Unmount the created partition
  6.  Detach the volume from VM instance and after that wait for volume to go to ‘Available’ state
  7.  Delete the volume
  8.  Delete the nova VM instance ==> It fails at this step, because status of server instance is set to Error

This happens one out of two times I run above test, so its not consistently seen. So wondering if someone has seen issue like this ?
BTW if I swap step 7) and 8), in other words delete nova VM instance before deleting the volume, then I do not see this issue and instance gets deleted correctly. So I am suspecting that either detach doesn’t completely detaches the volume from instance or deleting the instances some how triggers volume to detach indirectly.
So any idea, why and how this can happen ?

Thanks,
Harshil
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20140911/02ffbbb4/attachment.html>