This bug was found in with http://www.couchbase.com/issues/browse/MB-6007
Logs are attached on there.
Currently we're seeing rebalance to fail due to detected_nodes_change which purpose is to detect crashes of nodes that participate in rebalance.
|For Gerrit Dashboard: &For+MB-6280=message:MB-6280|
|23554,1||MB-6280 Stop rebalance only when important nodes go down.||ns_server||Status: ABANDONED||-1||0|