apparent deadlock in ep-engine (was: [system test] cluster crashed with error "exception exit: couldnt_connect_to_memcached")

Environment: Ubuntu 12.04 64 bit
1:10.3.3.239
2:10.3.3.199
3:10.3.3.215
4:10.3.3.240
5:10.3.2.85
6:10.3.3.218

Create 2 buckets: default and sasl. Each bucket is 1.2 GB
Create one doc with 2 views for each bucket.
Default bucket has 1 replica, replica index enable
Sasl bucket has 1 replica, replica index disable
Do the system test as in the following specification:
http://hub.internal.couchbase.com/confluence/display/QA/views+%28and+now+with+XDCR%29+tests

Do rebalance out node 10.3.3.218, rebalance failed
Then one by one, each node is going to unstable state (yellow)

Link to collect info of all nodes https://s3.amazonaws.com/packages.couchbase/collect_info/2_0_1/201304/6ubuntu_1204_build_210-185_cluster-unstable_20130408-160059.tgz

Link to manifest file of build 2.0.1-170 http://builds.hq.northscale.net/latestbuilds/couchbase-server-enterprise_x86_64_2.0.1-170-rel.deb.manifest.xml

On node 10.3.3.239, do kill -s SIGUSR1 1074 to generate erlang crash dump
Link to erlang crash dump https://s3.amazonaws.com/packages.couchbase/erlang/unix/erl_crash.dump.04-05-2013-node-3_239.gz

Cluster is currently in failed state.