[Update] Puppet failed on added controller node to the patched cluster

{"build_id": "2014-08-27_10-46-57", "ostf_sha": "4dcd99cc4bfa19f52d4b87ed321eb84ff03844da", "build_number": "481", "auth_required": true, "api": "1.0", "nailgun_sha": "04e3f9d9ad3140cd63a9b5a1a302c03ebe64fd0a", "production": "docker", "fuelmain_sha": "74a97d500bb2fe9528f99771ccc2ec657ae3f76e", "astute_sha": "bc60b7d027ab244039f48c505ac52ab8eb0a990c", "feature_groups": ["mirantis"], "release": "5.1", "fuellib_sha": "6f478caa9111e428b3c053eef44754f59ce7a2c2"}

Steps to reproduce:
1. Create ha cluster with 5.0-2014.1 (I use 5.0-26 iso for this) 1 controller +1 compute +1 cinder with nova network
2. Upgrade master node to 5.1 (I use 481 tarball for this)
3. Navigate to UI in action tab and run update
4. Ass soon as cluster ready after update add one/two nodes with controller role to the cluster and re-deploy changes

Expected:
Cluster ready. Ostf passed

Actual
Puppet failed on swift on added controller
http://paste.openstack.org/show/101041/

P.S cluster id 1, controller node
added controller node - node-8