remove run_filter_once_per_request from GroupAntiAffinityFilter

[Guangya Liu (Jay Lau)]

Now GroupAntiAffinityFilter was defined to run once per request, but this is not right, as filter_properties['group_hosts'] will be updated after one host was selected for one VM. So if there are multiple VMs in one request, the VM deployment will not obey GroupAntiAffinityFilter, as some VMs might be deployed on to same host with current VMs.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/39286

[Russell Bryant (russellb)]

Marked as Incomplete and put on the RC potential list since the patch is currently abandoned

[Guangya Liu (Jay Lau) (jay-lau-513)]

Hi Russell, it is going to be merged now: https://review.openstack.org/#/c/45571/

Seems launchpad did not catch the new patch, here I just mark the issue as "Fix Committed ". Thanks.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/45571
Committed: http://github.com/openstack/nova/commit/4233e16ba2f5bb85e93494b6eeb9ffff31dbdf30
Submitter: Jenkins
Branch: master

commit 4233e16ba2f5bb85e93494b6eeb9ffff31dbdf30
Author: Gary Kotton <email address hidden>
Date: Sun Sep 8 04:29:24 2013 -0700

    Ensure anti affinity scheduling works

    The query for the group in the system_metadata would not correctly
    filter the instances with the groups.

    In addition to this change 4a03071db23196c4ac1ec0d8ea163c46887d3b08
    ensured that the scheduling filter was only run once.
    In cases when provisioning fails and rescheduling
    takes place the filter must be invoked (if it is in the filter tool
    chain).

    Fixes bug: 1218878
    Fixes bug: 1206502

    Change-Id: I623c73635c36ffa038ae01476a25ba0c0863e536

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/grizzly
Review: https://review.openstack.org/55154