Memcached fails to close connection to the ejecting node when doing a swap rebalance and cause issues in deletion of the bucket on the node.

In the daily sanity, after doing a swap rebalance, the memcached on the ejected node fails to close connection to one of the threads and hence bucket deletion on the machine fails. This is causing subsequent connections attempts to the node to fail and causing failures in the daily build sanity tests.
http://server.jenkins.couchbase.com/job/build_sanity_matrix/DISTRO=centos7,TYPE=4node/2846/consoleFull - Test_4 is the swap rebalance test that is failing and causing the subsequent tests to fail as well.

http://server.jenkins.couchbase.com/job/build_sanity_matrix/DISTRO=centos7,TYPE=4node/2846/artifact/logs/testrunner-17-Mar-15_00-46-33/test_3/ : This has the logs for the test and nodes except the ejected node (172.23.107.184)
Logs for 172.23.107.184: https://s3.amazonaws.com/bugdb/jira/MB-23328/172.23.107.184.zip