yahoo-eng-team team mailing list archive

[Tao Li <litao3721@xxxxxxx>]

Public bug reported:

Description
===========

Evacauting an instance failed in destination host when spawing a new
instance, the libvirt driver will cleanup the disk in shared
storage(ceph, etc). so the instance losted the data in disk. In fact we
should ensure the user data not lost when evacuating a instance failed
and also resize，migrate operations。

so we should cleanup the disk just only when booting the instance in
shared storage。Other operations failed won‘t cleanup disks in shared
storage.

Steps to reproduce
==================

1 Booting an instance with root disk in shared storage such as ceph.
2 Power off the host which the instance is located in.
3 When the instance is spawning in destination host, and construt an exception such as
 “Timeout waiting for vif plugging callback for instance xxxxxxxxxxxxxxxx”
4 Then the instance is set to ERROR state。

Expected result
===============
The root disk of the instance in ceph is not removed。

Actual result
=============
The root disk of the instance in ceph is removed

** Affects: nova
     Importance: Undecided
     Assignee: Tao Li (eric-litao)
         Status: New

** Changed in: nova
     Assignee: (unassigned) => Tao Li (eric-litao)

-- 
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/1652259

Title:
  We should not delete shared storage disk when evacuating instance
  failed in destination host

Status in OpenStack Compute (nova):
  New

Bug description:
  Description
  ===========

  Evacauting an instance failed in destination host when spawing a new
  instance, the libvirt driver will cleanup the disk in shared
  storage(ceph, etc). so the instance losted the data in disk. In fact
  we should ensure the user data not lost when evacuating a instance
  failed and also resize，migrate operations。

  so we should cleanup the disk just only when booting the instance in
  shared storage。Other operations failed won‘t cleanup disks in shared
  storage.

  Steps to reproduce
  ==================

  1 Booting an instance with root disk in shared storage such as ceph.
  2 Power off the host which the instance is located in.
  3 When the instance is spawning in destination host, and construt an exception such as
   “Timeout waiting for vif plugging callback for instance xxxxxxxxxxxxxxxx”
  4 Then the instance is set to ERROR state。

  Expected result
  ===============
  The root disk of the instance in ceph is not removed。

  Actual result
  =============
  The root disk of the instance in ceph is removed

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1652259/+subscriptions