[R2.20.59] TSN HA: After stopping physical interface of vhost0 switch over is causing 17 min traffic loss

[chhandak]

On TSN HA setup if we stop the physical interface of Vhost0, it is taking very long time for switch over to happen.
Observed total 17min of traffic loss.

Steps followed
-----------------------
root@nodei10:~# ifconfig bond0 down >>>> Stopped physical interface

root@nodei10:~# contrail-status
== Contrail vRouter ==
supervisor-vrouter: active
contrail-tor-agent-5 active
contrail-tor-agent-7 active
contrail-tor-agent-8 active
contrail-vrouter-agent active
contrail-vrouter-nodemgr active >>>>>>Still status is active

root@nodei10:~# contrail-status
== Contrail vRouter ==
supervisor-vrouter: active
contrail-tor-agent-5 active
contrail-tor-agent-7 active
contrail-tor-agent-8 active
contrail-vrouter-agent active
contrail-vrouter-nodemgr active

root@nodei6:~# netstat -anp | grep 4321
tcp 0 0 0.0.0.0:4321 0.0.0.0:* LISTEN 22549/haproxy
tcp 0 15180 192.168.22.1:33348 192.168.22.5:4321 ESTABLISHED 22549/haproxy
tcp 0 1 192.168.22.1:42690 192.168.22.5:4321 SYN_SENT 22549/haproxy >>>> Ha proxy still trying to connect same node
tcp 0 0 192.168.22.1:34321 192.168.22.3:7000 ESTABLISHED 14963/java
tcp 0 0 192.168.22.210:5673 192.168.22.3:43211 ESTABLISHED 22549/haproxy
tcp 0 0 192.168.22.210:4321 192.168.11.1:63200 ESTABLISHED 22549/haproxy

[OpenContrail Admin (ci-admin-f)]

Review in progress for https://review.opencontrail.org/11967
Submitter: Hari Prasad Killi (<email address hidden>)

[OpenContrail Admin (ci-admin-f)]

Reviewed: https://review.opencontrail.org/11967
Committed: http://github.org/Juniper/contrail-fabric-utils/commit/81432ec82b949b530d91f119fb02766b1bc0d73b
Submitter: Zuul
Branch: R2.20

commit 81432ec82b949b530d91f119fb02766b1bc0d73b
Author: hari <email address hidden>
Date: Tue Jun 23 21:13:12 2015 +0530

Update HA Proxy config

Adding TCP keepalive option for tor-agent haproxy.
closes-bug: 1467593

Change-Id: I1000985a345492464ffd14cd9ba2812df9bf9fec

[OpenContrail Admin (ci-admin-f)]

Review in progress for https://review.opencontrail.org/12025
Submitter: Hari Prasad Killi (<email address hidden>)

[chhandak (chhandak)]

When Inactivity probe is set properly in OVSDB server of QFX we haven't hit the issue.

There is known QFX issue with inactivity probe. 1467909 [QFX] inactivity-probe-duration is not getting set in ovsdb-server

To avoid the bug 1467909 we can set we have to set ssl port and inactivity probe using same cli.

Sample:
set protocols ovsdb controller 192.168.22.210 inactivity-probe-duration 5000 protocol ssl port 4321

[OpenContrail Admin (ci-admin-f)]

Reviewed: https://review.opencontrail.org/12025
Committed: http://github.org/Juniper/contrail-fabric-utils/commit/a1cb73f3aa92221f804328a493f083eeebcfa8cb
Submitter: Zuul
Branch: R2.20

commit a1cb73f3aa92221f804328a493f083eeebcfa8cb
Author: hari <email address hidden>
Date: Thu Jun 25 12:33:28 2015 +0530

Reverting change to HA proxy config for TOR-Agent.

If tor-agent receives data from TOR regularly, it won't send keep alive.
HA proxy restarts in case this happens for 3 minutes, with the earlier change.
Removing this to avoid this problem.

Change-Id: I7c59210d5856eaf9724387b07ca5e757d148e89c
related-bug: 1467593

[OpenContrail Admin (ci-admin-f)]

Review in progress for https://review.opencontrail.org/12071
Submitter: Hari Prasad Killi (<email address hidden>)

[OpenContrail Admin (ci-admin-f)]

Reviewed: https://review.opencontrail.org/12071
Committed: http://github.org/Juniper/contrail-fabric-utils/commit/86b3b146156c157646090da5367bbfffb45b6ed7
Submitter: Zuul
Branch: master

commit 86b3b146156c157646090da5367bbfffb45b6ed7
Author: hari <email address hidden>
Date: Mon Jun 29 11:18:21 2015 +0530

Update HA Proxy config

Adding TCP keepalive option for tor-agent haproxy.

Change-Id: I2f78e78971d0cee021ca43b6c4aa248800258284
closes-bug: 1467593