rebalance fails while waiting for memcached on failed over nodes ( "wait_for_memcached_failed")

in order to repro :
http://10.1.5.184:8080/view/All/job/three-nodes-run-conf/
click build now . this job will run the failover test and publish the logs and diagnostics

steps to repro :
let's assume n1,n2,n3
1- create a bucket on n1
2- add n2 to n1. kill memcached on n2
3- wait 10 seconds
4- add n3 to n1 , kill memcached on 3
5- wait 10 seconds
6- call rebalance
7- rebalance succeeds
8- push in 100 keys into different vbuckets
9- wait for replication
10- invoke failover on n2,n3
11- call rebalance
rebalance will fail
Rebalance exited with reason

{wait_for_memcached_failed, "FailoverNodesTests-4a448bb0-d65f-443f-9cd6-54b958201561", ['ns_1@10.1.5.193','ns_1@10.1.5.178']}

logs : 10.1.5.192 is the master node