yahoo-eng-team team mailing list archive

[OpenStack Infra <1558532@xxxxxxxxxxxxxxxxxx>]

Reviewed:  https://review.openstack.org/299045
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=609adcd38cdcc8e1906d8b4f93a72b7fa2032817
Submitter: Jenkins
Branch:    master

commit 609adcd38cdcc8e1906d8b4f93a72b7fa2032817
Author: zte-hanrong <han.rong3@xxxxxxxxxx>
Date:   Tue Mar 29 20:19:01 2016 +0800

    Fix resize to same host failed using anti-affinity group
    
    ServerGroupAntiAffinityFilter selects host which is not in in the
    hosts of the members of the group. An instance does not have
    anti-affinity with itself.
    The current code has problem when select hosts, the instance's
    current host will be filtered out by ServerGroupAntiAffinityFilter.
    This can lead to failure when resize an instance with anti-affinity
    group to the same host.
    
    In this patch, ServerGroupAntiAffinityFite is modified.
    
    Change-Id: Ie016f59f5b98bb9c70b3e33556bd747f79fc77bd
    Closes-Bug:#1558532


** Changed in: nova
       Status: In Progress => Fix Released

-- 
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/1558532

Title:
  AntiAffinity policy problems with resize and migrate to same host

Status in OpenStack Compute (nova):
  Fix Released

Bug description:
  1. version
  kilo 2015.1.0

  2.Relevant log files:
  nova-scheduler.log
  （Log contains “========”  is added by myself）

  2016-03-17 17:54:23.092 15208 DEBUG nova.scheduler.filters.affinity_filter [req-3612987d-bef8-4727-811f-1dd668ac3328 eae626536b034af1a3635a36645dc54a 4d054614f49644c1935bdade9703c99b - - -] ================group_hosts: '[u'SBCJNailSlot3', u'sbcj-chenling-02-slot11']' host_passes /usr/lib/python2.7/site-packages/nova/scheduler/filters/affinity_filter.py:98
  2016-03-17 17:54:23.093 15208 DEBUG nova.scheduler.filters.affinity_filter [req-3612987d-bef8-4727-811f-1dd668ac3328 eae626536b034af1a3635a36645dc54a 4d054614f49644c1935bdade9703c99b - - -] Group anti affinity: check if sbcj-chenling-02-slot11 not in [u'SBCJNailSlot3', u'sbcj-chenling-02-slot11'] host_passes /usr/lib/python2.7/site-packages/nova/scheduler/filters/affinity_filter.py:101
  2016-03-17 17:54:23.094 15208 DEBUG nova.scheduler.filters.affinity_filter [req-3612987d-bef8-4727-811f-1dd668ac3328 eae626536b034af1a3635a36645dc54a 4d054614f49644c1935bdade9703c99b - - -] ================group_hosts: '[u'SBCJNailSlot3', u'sbcj-chenling-02-slot11']' host_passes /usr/lib/python2.7/site-packages/nova/scheduler/filters/affinity_filter.py:98
  2016-03-17 17:54:23.094 15208 DEBUG nova.scheduler.filters.affinity_filter [req-3612987d-bef8-4727-811f-1dd668ac3328 eae626536b034af1a3635a36645dc54a 4d054614f49644c1935bdade9703c99b - - -] Group anti affinity: check if SBCJNailSlot3 not in [u'SBCJNailSlot3', u'sbcj-chenling-02-slot11'] host_passes /usr/lib/python2.7/site-packages/nova/scheduler/filters/affinity_filter.py:101
  2016-03-17 17:54:23.094 15208 INFO nova.filters [req-3612987d-bef8-4727-811f-1dd668ac3328 eae626536b034af1a3635a36645dc54a 4d054614f49644c1935bdade9703c99b - - -] Filter ServerGroupAntiAffinityFilter returned 0 hosts for instance c48ee7d6-451e-4b96-ac39-261082f8950e
  2016-03-17 17:54:23.095 15208 DEBUG nova.scheduler.filter_scheduler [req-3612987d-bef8-4727-811f-1dd668ac3328 eae626536b034af1a3635a36645dc54a 4d054614f49644c1935bdade9703c99b - - -] There are 0 hosts available but 1 instances requested to build. select_destinations /usr/lib/python2.7/site-packages/nova/scheduler/filter_scheduler.py:77

  3.Reproduce steps:
  *3.1
  create a anti-affinity group with policy 'anti-affinity' named "antiaffinitygroup"
  boot two vms using "antiaffinitygroup",named "test_wzz1" and "test_wzz2"

  results:
  [root@sbcj-chenling-02-slot11 ~(keystone_admin)]# nova list
  +--------------------------------------+-----------+--------+------------+-------------+-------------------+
  | ID                                   | Name      | Status | Task State | Power State | Networks          |
  +--------------------------------------+-----------+--------+------------+-------------+-------------------+
  | c48ee7d6-451e-4b96-ac39-261082f8950e | test_wzz1 | ACTIVE | -          | Running     | test0301=3.3.3.60 |
  | 55107de8-38bc-4b3a-8681-df17746e4f11 | test_wzz2 | ACTIVE | -          | Running     | test0301=3.3.3.61 |
  +--------------------------------------+-----------+--------+------------+-------------+-------------------+

  [root@sbcj-chenling-02-slot11 ~(keystone_admin)]# nova server-group-list
  +--------------------------------------+-------------------+--------------------+------------------------------------------------------------------------------------+----------+
  | Id                                   | Name              | Policies           | Members                                                                            | Metadata |
  +--------------------------------------+-------------------+--------------------+------------------------------------------------------------------------------------+----------+
  | 79ebfbfc-6e7c-4bf0-97c5-14d7fca22e96 | antiaffinitygroup | [u'anti-affinity'] | [u'55107de8-38bc-4b3a-8681-df17746e4f11', u'c48ee7d6-451e-4b96-ac39-261082f8950e'] | {}       |
  +--------------------------------------+-------------------+--------------------+------------------------------------------------------------------------------------+----------+

  *3.2
  I have two compute nodes：sbcj-chenling-02-slot11，SBCJNailSlot3.
  The vm test_wzz1 in SBCJNailSlot3,and the test_wzz2 in sbcj-chenling-02-slot11,because of the anti-affinity group
   Then set allow_resize_to_same_host=true,and add ServerGroupAntiAffinityFilter in scheduler_default_filters in nova.conf,and restart services.

  *3.3
   now，execute the command : nova resize test_wzz1 m1.small

  Example：
  [root@sbcj-chenling-02-slot11 ~(keystone_admin)]# nova resize test_wzz1 m1.small

  Expected result:
  vm resize to the same host successfully.

  Actual result:
  ERROR (BadRequest): No valid host was found. No valid host found for resize: No valid host was found. There are not enough hosts available. (HTTP 400) (Request-ID: req-3612987d-bef8-4727-811f-1dd668ac3328)
  Relevant log files: See section 2

  4.Bonus points for reproducing shell script / test:
  Operation cold-migration has the same problem.
  It seems  test_wzz1 anti-affinity against to itself,so resize failed although we allow vm resize to same host.
  This is unreasonable.

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1558532/+subscriptions