Some Vms failed to get IPs in Scale run

[Aman Kumar]

In scale run, we spawned 1000 VMs across 20 networks .

Out of 1000 VMs , 993 VMs came to active state .
One of 993 VM failed to get IP.

When we keep spawning VMs, there will be constant
security group member update calls, which results in firewall refreshes
and clears openflows rules belonging to VMs in other hosts.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/249146

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/249146
Committed: https://git.openstack.org/cgit/openstack/networking-vsphere/commit/?id=faf6b8b5e68a2a2c2cc6065abd077f62168cca94
Submitter: Jenkins
Branch: master

commit faf6b8b5e68a2a2c2cc6065abd077f62168cca94
Author: Aman Kumar <email address hidden>
Date: Tue Nov 24 03:30:57 2015 -0800

    Decrease the void of flow existence

    In scale tests, as we keep spawning VMs, there will be constant
    security group member update calls, which results in firewall refreshes.
    Existing implementation clears openflows rules belonging to VMs in other
    hosts. When the scale run is in progress, we see DRS kicking in and moving
    VMs to other hosts during their poweron. Since we are clearing the rules at
    times, its resulting in VMs not getting IPs.
    This patch addresses the issue by not clearing the provider rules, which will
    help the VMs obtain IP address.

    Co-Authored-By: Sudhakar Babu Gariganti <email address hidden>

    Change-Id: I7d01ba6f3a40c7b00aec484479f93237cfbb2f97
    Closes-Bug: #1519219

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/liberty
Review: https://review.openstack.org/254044

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/254044
Committed: https://git.openstack.org/cgit/openstack/networking-vsphere/commit/?id=a4853aa161b680272796afe41e06d86d7e33c5fd
Submitter: Jenkins
Branch: stable/liberty

commit a4853aa161b680272796afe41e06d86d7e33c5fd
Author: Aman Kumar <email address hidden>
Date: Tue Nov 24 03:30:57 2015 -0800

    Decrease the void of flow existence

    In scale tests, as we keep spawning VMs, there will be constant
    security group member update calls, which results in firewall refreshes.
    Existing implementation clears openflows rules belonging to VMs in other
    hosts. When the scale run is in progress, we see DRS kicking in and moving
    VMs to other hosts during their poweron. Since we are clearing the rules at
    times, its resulting in VMs not getting IPs.
    This patch addresses the issue by not clearing the provider rules, which will
    help the VMs obtain IP address.

    (cherry picked from commit faf6b8b5e68a2a2c2cc6065abd077f62168cca94)

    Co-Authored-By: Sudhakar Babu Gariganti <email address hidden>

    Change-Id: I7d01ba6f3a40c7b00aec484479f93237cfbb2f97
    Closes-Bug: #1519219