yahoo-eng-team team mailing list archive

[Balazs Gibizer <balazs.gibizer@xxxxxxxx>]

This is more a libvirt / nfs question than a nova question. I suggest to
reach out to the libvirt community: https://libvirt.org/bugs.html

** Changed in: nova
       Status: New => Invalid

-- 
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/1797333

Title:
  Instances are locked and unable to start after server crash (queens)

Status in OpenStack Compute (nova):
  Invalid

Bug description:
  After restarting crashed host, disks of hosted instances on nfs are
  locked and cannot be restarted:

  libvirtError: internal error: process exited while connecting to monitor: 2018-10-10T10:16:09.816477Z qemu-system-x86_64: -drive file=/var/lib/nova/instances/ed7760a8-3008-4feb-83f3-3b753b0e7d6e/disk,format=qcow2,if=none,id=drive-virtio-disk0,cache=none: Failed to get "write" lock
  ERROR nova.compute.manager [instance: ed7760a8-3008-4feb-83f3-3b753b0e7d6e] Is another process using the image?

  The same situation occurs on other compute nodes connected to the same
  shared file system, after evacuate instances. So it seems that disks
  are locked by libvirt in an unknown, undocumented way. As workaround I
  had to make copy of all failed instances, delete their disk files and
  restore them from copy. After that instances started successfully.

  If there is other solution to unlock those instance disks, please
  share.

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1797333/+subscriptions