Comment 0 for bug 1475296

When primary controller was unaccessible or deleted from cluster master mongo moved to the other controller.
In this case primary-mongo role and master host for replicaset of mongoDB are placed on the different controller nodes puppet fails

The possible solution is update mongoDB from upstream:
https://github.com/puppetlabs/puppetlabs-mongodb/blob/master/lib/puppet/provider/mongodb_replset/mongo.rb#L212-L217

Reproduced in fuel 6.1 and 7.0

Step to reproduce:
1) Build cluster using controller+mongodb
2) Delete primary-controller.
3) Add new controller to cluster.

The last tested version of fuel:
 release: "7.0"
 openstack_version: "2014.2.2-7.0"
 api: "1.0"
 build_number: "141"
 build_id: "2015-06-30_19-02-39"
 nailgun_sha: "9ea8b2cf30966b4805d0655b516c5a5d67d05865"
 python-fuelclient_sha: "7f50043963bda38b7857cfcb61ece2bcc5b9aa1c"
 astute_sha: "8d11177f91aab90f8034e9f6a479f343e1a4ec7f"
 fuel-library_sha: "21131baa6dbe2f79fc14cdc2667c301f0be97a9c"
 fuel-ostf_sha: "2eff8e314d3024707f32cf539a7997bfc427f351"
 fuelmain_sha: "b6f98e6b31cee7f9b925c6cc24e47966fa63e8ec"