LB changes like healthmonitor, member addition/deletion does not update existing netns SI

[Vedamurthy Joshi]

R1.10 LB Build 8

Created a LB , 2 backend members and a VIP.
Client was able to see loadbalanced HTTP traffic.
Create a healthmonitor to detect the downtime in 10 sec.. and assign it to the LB.
Bring down one of the backend members.
Client requests continue to reach the VM which is down.

If we create the netns SI after the healthmonitor config is applied to the pool, then lb-members going down is detected fine .

root@nodeh5:~# vi /var/lib/contrail/loadbalancer/2a4deae6-430c-4479-823d-b9a417fa6822/etc/haproxy/haproxy.cfg
global
    daemon
    user nobody
    group nogroup

defaults
    log global
    retries 3
    option redispatch
    timeout connect 5000
    timeout client 50000
    timeout server 50000

frontend e28e6220-05c8-4880-85a7-2c6efd7eb644
    bind 101.1.1.3:80
    mode http
    default_backend 2a4deae6-430c-4479-823d-b9a417fa6822

backend 2a4deae6-430c-4479-823d-b9a417fa6822
    mode http
    balance roundrobin
    server 5a63f809-9aff-4d41-8e93-e9fa65e29de3 100.1.1.3:80 weight 1 check inter 5000 rise 1 fall 2
    server 6c12570b-1667-4fcd-a778-5c4e5595816e 100.1.1.2:80 weight 1 check inter 5000 rise 1 fall 2

[Vedamurthy Joshi (vedujoshi)]

Divakar already fixed it...before the merge to R1.10 branch

[Divakar Dharanalakota (ddivakar)]

Change ID afa92c0b4c1e7555f717fda609505189efc1cb7f has fix for this in mainline.
-Divakar