View fails when rebalancing from 3 to 1 node and back up.
During the whole process, access a view.
3 server in one cluster.
- Manual failover one node, and then rebalance. It is fine.
- Then manual failover another one, and then rebalance, and at this time in the system there is only one node,
- Add 2 nodes back into the cluster and rebalance.
- Sometimes it will throw the "Failed to access the view" exception, and sometimes the CouchbaseClient.getView() call will return NULL.
This issue is also seen when adding nodes to a cluster.