yahoo-eng-team team mailing list archive

[alex <chinabjalex@xxxxxxxxx>]

Public bug reported:

Description
===========
Normally, if you finished evacuating instance, nova-compute will destroy the existed instances whose state are "done" or "accept" in the migration table from the fault host after the nova-compute service is restarted. However, if the nova-scheduler failed to schedule the hosts, e.g. no valid host, nova-scheduler won't update the migration state of the instance to "failed", so that the nova-compute will destroy this instance after restarting by mistake.

Steps to reproduce
==================
1. Create some instances in the the specific host.
2. Make this host to fault state.
3. Disable nova-compute service in other hosts, aim to mock that nova-scheduler fail to schedule.
4. Recover the fault host and restart nova-compute.
5. Check all instances are still existed in the kvm.

Expected result
===============
All instances should be still existed in the kvm.

Actual result
=============
All instances are destroyed by nova-compute unfortunately.

** Affects: nova
     Importance: Undecided
     Assignee: alex (chinabjalex)
         Status: New

** Changed in: nova
     Assignee: (unassigned) => alex (chinabjalex)

-- 
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/1583504

Title:
  The instances which didn't be evacuated will be destroyed when the
  nova-compute service is restarted.

Status in OpenStack Compute (nova):
  New

Bug description:
  Description
  ===========
  Normally, if you finished evacuating instance, nova-compute will destroy the existed instances whose state are "done" or "accept" in the migration table from the fault host after the nova-compute service is restarted. However, if the nova-scheduler failed to schedule the hosts, e.g. no valid host, nova-scheduler won't update the migration state of the instance to "failed", so that the nova-compute will destroy this instance after restarting by mistake.

  Steps to reproduce
  ==================
  1. Create some instances in the the specific host.
  2. Make this host to fault state.
  3. Disable nova-compute service in other hosts, aim to mock that nova-scheduler fail to schedule.
  4. Recover the fault host and restart nova-compute.
  5. Check all instances are still existed in the kvm.

  Expected result
  ===============
  All instances should be still existed in the kvm.

  Actual result
  =============
  All instances are destroyed by nova-compute unfortunately.

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1583504/+subscriptions