Occasional cluster failure due to global-instruction-queue

It seems there is a component not migrated to singleton yet, so booting up several cluster members at one has a chance to fail the Config SubSystem (server is unhealthy).

This Bug causes an occasional red dot in Carbon 3node csit all jobs.

See this [0] log, the first sign is:

2017-04-10 11:20:17,635 | WARN | ult-dispatcher-3 | ShardDataTree | 187 - org.opendaylight.controller.sal-distributed-datastore - 1.5.0.SNAPSHOT | member-2-shard-default-config: Store Tx member-2-datastore-config-fe-0-txn-33-0: Conflicting modification for path /(urn:opendaylight:params:xml:ns:yang:odl:programming:config?revision=2017-03-01)odl-programming/odl-programming-config/odl-programming-config[

{(urn:opendaylight:params:xml:ns:yang:odl:programming:config?revision=2017-03-01)instruction-queue-id=global-instruction-queue}

].

[0] https://logs.opendaylight.org/releng/jenkins092/controller-csit-3node-clustering-all-carbon/268/console.log.gz