Comment 3 for bug 1312666

First, the traceback provided isn't related to this bug. Second, if node is marked on UI as 'error' - this is a proper behaviour. Cluster may be even operational, but in this particular case one of compute nodes failed, so the whole deployment was marked as failed on UI. The thing is, the failed node was primary for MongoDB, so it would break something Ceilometer-related, anyway.