Recovery rebalance fails after failover

My guess there is some issue in ep-engine while deleting the bucket after failover.

Repro Steps:

1. 3 node cluster with 1 bucket
2. Hard failover Node1
3. Add back Node 1 - either using delta or full recovery.
4. Rebalance

Rebalance fails.

If it is delta recovery rebalance, rebalance fails due to timeout while trying to retrieve warmup information from memcached.

If it full recovery rebalance, failure is due to old_buckets_shutdown_wait_failed.

But, based on the memcached logs, root cause looks the same. There seems to be some issue with delete bucket after failover.

Easy to repro on latest Vulcan but I have attached cbcollect files.