Summary : When the backup is trying to sync with the master and the master stops (due to paging disk missing) the backup will then try to become live. As expected - it fails as it has not sync'd fully with the original live server. Problem: The broker appears to stay in this state it does not shutdown or go back to backup. I would expect it to either to shutdown or restart as a backup broker.

Detail description:

• if both live (master broker) and backup (slave broker) have been stopped.

• the master broker configured to be "live" is started and the paging directory (disk) is not there, it can still be started as live.

• the slave broker is started and tries to sync with it and "live" master broker realises it has no paging directory and shuts down. (as expected)

• the backup detects the live has shutdown and is being promoted to live. It then hits the following exception and does not appear to go live.

• the slave broker detects the "live" has gone tries to go to "live" and throws the following exception

2018-08-09 15:26:56,369 ERROR [org.apache.activemq.artemis.core.server] AMQ224000: Failure in initialisation: ActiveMQIllegalStateException[errorType=ILLEGAL_STATE message=AMQ119026: Backup Server was not yet in sync with live]
	at org.apache.activemq.artemis.core.server.impl.SharedNothingBackupActivation.run(SharedNothingBackupActivation.java:316) [artemis-server-2.6.1.amq-720005.jar:2.6.1.amq-720005]
	at org.apache.activemq.artemis.core.server.impl.ActiveMQServerImpl$ActivationThread.run(ActiveMQServerImpl.java:3130) [artemis-server-2.6.1.amq-720005.jar:2.6.1.amq-720005]

• if the master broker is restarted (with the disk back in place) the master goes to live but the slave broker does not appear to go to backup. It appears to remain in that state.

I would have expected the broker to restart as a backup broker.