[R5.1 k8s]: In multi controller cluster, Kube manager restart takes 2 minutes to elect the active kube manager.

[Pulkit Tandon]

R4.0-ocata-156-ubuntu16

3 Controllers and 1 Kube master
2 Agents and Kube slaves
Multi node HA setup

Description:
Restart the contrail-kube-manager on all 3 controller nodes.
After restart, the election of active kube manager takes around 2 minutes.
By that time, if any request to delete Pod is made, contrail-kube-manager misses that request and the POD VM entry get stuck in the agent for close to 3 minutes.

Logs:
Attached are contrail-kube-manager and kube-manager-zk logs for all 3 controller nodes.
Note that in the logs, the restart was done at almost:
04:08:42
Logs start from that point and end till the last entry logged.

[Vineet Gupta (vineetrf)]

This bug can be tested once the below bug is fixed.

https://bugs.launchpad.net/juniperopenstack/+bug/1749147

[Pulkit Tandon (pulkitt)]

Verified this issue on R5.0-45-ocata.
The provisioning was micro services. The issue was not observed on microservicess setup.
Hence closing the issue on R5.0 branch

[Pulkit Tandon (pulkitt)]

Any reason of marking it "won't fix" for R4.1 ?