Create 3 server groups and 9 pods, 3 in each group. Kill a pod from each group. Cluster has 1 bucket with 1 replica, Autofailover timeout is 10 seocnds. Expected behaviour is for one pod to be failed over. Once the pod fails over, the test resets autofailover counter. Then the next pod is expected to failover, followed by a autofailover counter reset. Finally the last pod is failed over and the cluster is back to 9 healthy nodes. What I see in the tests is the failovers never take place.