This issue was resolved in OneFS 8.0.0.4.   See the release notes for the Resolved Issue.   During failover to a secondary ResourceManager, HDFS MapReduce jobs might have been disrupted. This could have occurred because, during failover, OneFS renegotiated the connection to the ResourceManager using the same Kerberos ticket but with a different name. As a result, the request to connect to the secondary ResourceManager could not be authenticated and access was denied.181448 ... View more