ns_janitor may lose replicas of nearly completed vbucket moves (was: {DCP} : Delta Recovery Impossible after re-try of graceful failover since in first attempt failed)

build 1186

Scenario
1. Create a 7 Node cluster
2. Create default bucket with 200 K items
3. Graceful failover a node
4. Kill memcached of another node during graceful failover
5. Graceful failover the same node in step 3
6. Add-back the node with delta recovery
7. Hit Rebalance

We see the following warning:: "Fail Over Warning: Rebalance required, some data is not currently replicated!”

In Step 7, Rebalance fails for delta recovery. Says delta recovery is not possible. Although we see nodes in the cluster are in healthy state. This is true when we have 200 K items Vs 100K items where it passes.

I am attaching the logs for you to analyze. Since the above warning comes in both cases. Not sure about the internal state of the system which stops the add-back delta recovery.

Test fails for 2k items
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.144-8222014-1436-diag.zip
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.144-8222014-1445-couch.tar.gz
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.145-8222014-1437-diag.zip
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.145-8222014-1445-couch.tar.gz
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.146-8222014-1439-diag.zip
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.146-8222014-1445-couch.tar.gz
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.147-8222014-1440-diag.zip
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.147-8222014-1446-couch.tar.gz
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.148-8222014-1441-diag.zip
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.148-8222014-1446-couch.tar.gz
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.149-8222014-1442-diag.zip
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.149-8222014-1446-couch.tar.gz
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.150-8222014-1444-diag.zip
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.150-8222014-1446-couch.tar.gz

Test passes for 1 K Items

https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.144-8222014-1458-diag.zip
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.144-8222014-159-couch.tar.gz
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.145-8222014-150-diag.zip
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.145-8222014-159-couch.tar.gz
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.146-8222014-151-diag.zip
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.146-8222014-159-couch.tar.gz
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.147-8222014-153-diag.zip
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.147-8222014-159-couch.tar.gz
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.148-8222014-1510-couch.tar.gz
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.148-8222014-154-diag.zip
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.149-8222014-1510-couch.tar.gz
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.149-8222014-156-diag.zip
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.150-8222014-1510-couch.tar.gz
https://s3.amazonaws.com/bugdb/jira/MB-12037/10.6.2.150-8222014-157-diag.zip