Checks to do incase of unexpected failover between VMs hosted in compute based on contrail.

Unexpected HA failover of tenant VMs

The first thing to identify is whether customer uses Allowed Address Pair , which again can be of two types one with virtual mac (eg VRRP) and one without virtual mac (eg PCS). Refer KB https://supportportal.juniper.net/s/article/Contrail-How-to-configure-Allowed-Address-Pair-with-without-virtual-MAC-address to identify your use case.

Incase of unexpected fail over, It is recommended to look into logs of HA application running inside the tenant VMs.

See if there are any missing heartbeats , liveness check timeouts etc

Based on the issue seen, troubleshooting approach may differ.

However If incase of failover, consider following data points.

+ Access or console to tenant VMs and identify where the virtual address ( VIP / Floating IP ) resides

+ Start a capture for all participating interfaces of HA application.

+ Check if you are able to ping Virtual IP and packets are received in active host or wherever IP resides.

+ Check if Gracious ARP is sent out in the event of switchover / failover, this will perform route updation otherwise packets may still route to inactive interface.