RabbitMQ returns error: 'Error: {aborted,{no_exists,[rabbit_user,{internal_user,'_','_','_'}]}}'

[Artem Panchenko]

api: '1.0'
astute_sha: 277b694bc6c2aa68601f0be96b6ab94a0cf6fe37
auth_required: true
build_id: 2014-11-26_23-06-29
build_number: '12'
feature_groups:
- mirantis
fuellib_sha: 27e4449ad4c68a35afdd9d281951cf43b39a137f
fuelmain_sha: 1a784744731976f43ff2fc88e678a3fea35b8bdc
nailgun_sha: c25bfbe204981572296ae62d78b6ce5bf7de2f2a
ostf_sha: a35f516f1606b0d03d51ff63bfe3fbe23de4b622
production: docker
release: '6.0'

Deployment of cloud failed due to errors in puppet run on one of controllers:

http://paste.openstack.org/show/139448/

Steps to reproduce:

1. Create new environment: Ubuntu, HA, NeutronVlan, Ceph volumes&images
2. Add nodes: 3 controller+ceph and 2 compute+ceph
3. Deploy changes

Expected result:

- new cloud is successfully deployed

Actual:

- "Deployment has failed. Check these nodes: 'slave-02_controller_ceph-osd'"

Seems that on one of controller nodes (node-3) had broken RabbitMQ service, according to its status it was not joined to cluster:

http://paste.openstack.org/show/139456/

but 'crm status' said that everything is okay:

http://paste.openstack.org/show/139465/

After I joined it to cluster manually 'list_users' command became to work fine:

http://paste.openstack.org/show/139472/

Diagnostic snapshot is attached.