I have a 3 node master setup for marathon and mesos masters, everything is fine until I reboot one of the nodes or try restarting zookeeper.
The issue is somehow correlating to the internal DB zookeeper is using, when I delete /var/lib/zookeeper/version2/* then zookeeper comes up fine and re syncs with the other nodes.
I am using the current marathon mesos packages with zookeeper from the mesosphere repo on rhel7.
Does anybody know how to fix this? The Filesystem below does not change anything, I tried with xfs and ext4.
Another question would be to somehow backup the marathon apps - if I reboot let's say all nodes in the quorum with deleting the zookeeper DB everything is lost.
