failover of marathon to host not running mesos-master

Configure one host1 running:

• marathon
• mesos-master
• zookeeper

Configure second host (192.168.56.11) running:

• marathon

I'm running these on centos 7 as services.

I configured host1 /etc/mesos-master and /etc/mesos-slave with hostname and ip set to host 1 ip. And I set /etc/zk to point to host 1 zk. Host1 marathon works, deploys apps, etc.

I configured host2 /etc/marathon/conf/hostname to host 2 ip, carefully set master and zk to the appropriate host1 addresses, and added file /etc/default/marathon/LIBPROCESS_IP with contents of the host2 ip.

Upon starting, host1 is leader. Stopping marathon on host1 does not failover to host2. visiting host2 :8080 "about" says host2 is leader, but host1:5050 shows host2 as an "inactive framework". journalctl on host 2 has message for marathon:

"Scheduler driver bound to loopback ..... you might want to set 'LIBPROCESS_IP' environment variable ..."

What am I missing? I am trying to achieve/observe marathon failover.