Analytics service is unavailable if network goes down for 30s on NC node.

1. There is 1 KV node in the cluster, a default bucket with 100k docs.
2. Add 2 cbas(CC+NC) node and do the initial setup.
3. Queue 100 queries on cbas: "select sleep(count,50000) from default_ds;"
4. Load many more 200k docs to make sure ingestion is in progress.
5. Block Network NC node for 30s : ifdown -a && sleep 30 && ifup -a (ubuntu)
6. After network comes up on NC node try to access workbench. It throws 504 error.
http://10.142.160.103:8095/analytics/service is unreachable
7. Rebalance out NC node and observe that rebalance stuck at a point.