I’m trying to use Karaf HA/failover feature documented here : http://karaf.apache.org/manual/latest/#_ha_failover_and_cluster
With database lock, I see that Karaf running with the Lock as Active instance is just shutting down upon losing connection with DB.
I think, it would better if Karaf tries few times to reconnect & acquire lock before doing shutdown. When only one instance of karaf is running and DB connection fluctuates, this behavior impacts availability (needs restart from orchestration layer).
Note: I'm using start level = 1 and karaf.lock.slave.block = true