We noticed this only happens when failing over from the CAS configured as secondary to the primary CAS. The client still appears as certified but has no longer access to the network for around 2 minutes.
When we failover from the primary to the secondary CAS the client stays connected without losing connectivity.
We configured another pair today and we are noticing the same behaviour, however it seems random... sometimes the user barely looses connection, other times it will take from 2-5 minutes for it to come back.
We are only using eth2 for the failover link since we only have one serial port.
When we test we make sure both servers are up and then we reboot the primary. The secondary becomes active immediately. When both are up again we repeat the process.