This looks to be an intermittent failure in cluster bring up.

If this is a random failure appreciate if you could remove the failing tests or stabilize them. 

We need to reach a point where this subset always passes as a basis for a gate.

Problem is definitely outside of daexim.

Keyword "ClusterManagement .Start_Members_From_List_Or_All" fail to reach last node in cluster.

There is nothing we can do about it other then disable failing test (but that will only hide real issue).

Quick look at https://logs.opendaylight.org/releng/vex-yul-odl-jenkins-1/daexim-csit-3node-clustering-basic-only-fluorine/249/odl_3/odl3_karaf.log.gz

shows:

2018-10-23T03:06:37,040 | ERROR | opendaylight-cluster-data-akka.actor.default-dispatcher-40 | ActorSystemImpl                  | 43 - com.typesafe.akka.slf4j - 2.5.11 | Uncaught error from thread [opendaylight-cluster-data-akka.actor.default-dispatcher-32]: org/w3c/dom/ElementTraversal, shutting down JVM since 'akka.jvm-exit-on-fatal-error' is enabled for ActorSystem[opendaylight-cluster-data]
java.lang.NoClassDefFoundError: org/w3c/dom/ElementTraversal

which is similar to https://jira.opendaylight.org/browse/NETCONF-576

In any case, we can't fix that in daexim.

There is nothing to be done in daexim and problem is already tracked in CONTROLLER-1867

Appreciate the detailed answer.

Will raise it today on the TSC call to decide how to progress.

That fact that it's the same problem as in the netconf is a very important point.