yahoo-eng-team team mailing list archive

[Ian Cordasco <icordasc+launchpad@xxxxxxxxxx>]

So most services in openstack (with the exception of Nova) don't tend to
know anything about the services consuming them. Cinder doesn't know if
an instance still depends on a volume, we don't know if an instance
still relies on an image. In fact, I don't see a reason why we should be
aware of that. Perhaps what is actually desired is a way for an owner to
prevent an image from being deleted but I'm not sure we want yet more
logic around that and I'm not quite certain there's a great deal of
value in that.

** Changed in: glance
       Status: New => Opinion

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to Glance.
https://bugs.launchpad.net/bugs/1485635

Title:
  Delete image should not progress as long as instances using it

Status in Glance:
  Opinion

Bug description:
  Currently it is possible to delete the glance images, even if instances are still referencing it.
  This is in particular an issue once you deleted the image and tried to resize/migrate instances to a new host.
  The new host can't download the image from glance and the instance can't start anymore due to the missing qemu backing file. In case of a resize, the action does abort since the coalescing of the qemu file, happening during resize,  can not be completed without the qemu backing file. 
  This issue usually causes manual intervention to reset the instance action, state and manual search/copy of the base image file to the target host.
  The ideal state would be to prevent image-delete requests as long as active instances are still referencing it.

To manage notifications about this bug go to:
https://bugs.launchpad.net/glance/+bug/1485635/+subscriptions