Rebalance failed while doc ops are running and KV swap with cbas rebalance in is performed on the cluster with error "The stream closed early because the conn was disconnected".

Here are the actual steps followed: https://docs.google.com/document/d/1I7yOqAAlRteWmbwS5fpJjqXCcW4n8JmK6fqYGUrKeqg/edit.
The test is ran at a much lower scale than mentioned in the document. Consider 10M -> 100 docs and 1M update/delete as 10 docs update/delete.
It is failing at step 50.

To make it simpler:
1. There are 3 buckets in the cluster.
2. Doc ops are performed only in 1 bucket "GleamBookUsers".
3. There are analytics nodes present in the cluster and cbas bucket and datasets are present.
4. Make sure create doc ops in progress and do a KV swap rebalance and CBAS node rebalance in in a single go.
5. Observe that rebalance will fail.

Here are the logs:
https://s3.amazonaws.com/bugdb/jira/rebalance_failed/collectinfo-2018-04-18T103925-ns_1%4010.142.164.101.zip
https://s3.amazonaws.com/bugdb/jira/rebalance_failed/collectinfo-2018-04-18T103925-ns_1%4010.142.164.102.zip
https://s3.amazonaws.com/bugdb/jira/rebalance_failed/collectinfo-2018-04-18T103925-ns_1%4010.142.164.103.zip
https://s3.amazonaws.com/bugdb/jira/rebalance_failed/collectinfo-2018-04-18T103925-ns_1%4010.142.164.104.zip
https://s3.amazonaws.com/bugdb/jira/rebalance_failed/collectinfo-2018-04-18T103925-ns_1%4010.142.164.105.zip
https://s3.amazonaws.com/bugdb/jira/rebalance_failed/collectinfo-2018-04-18T103925-ns_1%4010.142.164.106.zip

This issue is consistently reproducible on 5.5.0-2432 build.