Deleting and readding the same VIP results in a non-working VIP

[Liping Mao]

Environment:
Openstack Version : Havana
OS : CentOS6.3
LBaaS Driver : haproxy

Reproduce Steps:
1. In loadbalance pool, add vip :100.100.100.100. This vip is on a provider network and will use external Physical Router.
2. Access 100.100.100.100, make sure it can work.
3. Delete vip.
4. Add vip 100.100.100.100 again.
We will find that we can't access 100.100.100.100 at the second time.

[Liping Mao (limao)]

Route Cause:
When we access 100.100.100.100 at first time, there will be arp info in the External Router. And after we delete vip and create vip 100.100.100.100 again, the MAC change. But there still be ARP cache in external Router.

Solution:
When we create the port, we need send gratuitous ARP to flash the ARP cache in External Router.
Our deploy is on CentOS6.3, there is the following Bug make arping can't work in netns, so I think we can just use ping gateway to reflash the ARP cache on External Router:
https://bugzilla.redhat.com/show_bug.cgi?id=963927

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/84647

[Liping Mao (limao)]

arping can support netns in CentOS6.3 . and we need upgrade iputils.

https://rhn.redhat.com/errata/RHBA-2013-0901.html

[Liping Mao (limao)]

Any idea about this bug?

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/84647
Committed: https://git.openstack.org/cgit/openstack/neutron/commit/?id=e7d7414548a29c6b362f5e890cd8f8cb31885b84
Submitter: Jenkins
Branch: master

commit e7d7414548a29c6b362f5e890cd8f8cb31885b84
Author: LipingMao <email address hidden>
Date: Wed Apr 2 07:34:31 2014 +0000

    LBaaS VIP doesn't work after delete and re-add

    When delete and re-add the same vip, we need to send gratuitous ARP
    to flush the ARP cache in the Router.

    Change-Id: Id97088abb95f4433a100abdae8c8726b3be42ed2
    Closes-Bug: 1301165