This is a kind of general problem, because more than one suite enters the state when a node starts to respond with http response 401 (unauthorized). Few bugs may already be opened for problems related to the node isolation, but these problems result in odl node response 401.
At the moment i am unaware of how to recover from this state, but it is definitely possible, because every next suite behaves well at the start.

Job: https://jenkins.opendaylight.org/releng/view/controller/job/controller-csit-3node-clustering-only-carbon/626/

Suite: Rpc Provider Partition And Heal

• 2 nodes have rpc registered, the 3rd one no
• the one without rpc is isolated and we expect status code 501 when rpc called, but 401 is returned

Suite: Action Provider Partition And Heal
A node with routed rpc is isolated. The rpc is then called on that node and status 200 is expected. 401 is returned.

Suite: Partition And Heal (Clustering Singleton)
Singleton constant is installed on each node and then the owner node is isolated. When requesting odl-mdsal-lowlevel-target:get-singleton-constant on isolated node, 501 is expected, 401 is returned.

This may not be the single problem, but all is related to node isolation process.