AndyTomasello on "Issue getting agents to connect properly"

Hello! I'm having trouble to get my agents to properly connect. The OpsCenter server is sitting in my office. I have a Cassandra cluster sitting in two data centers. The broadcast address is being used on the Cassandra side. The broadcast as a public IP address but since it's NAT, they can't hit that address. I've configured:

local_interface
agent_rpc_interface
agent_rpc_broadcast_address
jmx_host

and I finally get OpsCenter to say the agent is connected. The only errors I'm still getting from the agent are:

ERROR [Thread-4] 2013-08-16 12:55:22,410 Unable to connect to Cassandra #<HectorException me.prettyprint.hector.api.exceptions.HectorException: All host pools marked down. Retry burden pushed out to client.>
WARN [Hector.me.prettyprint.cassandra.connection.CassandraHostRetryService-1] 2013-08-16 13:07:01,040 Downed 208.49.103.92(208.49.103.92):9160 host still appears to be down: Unable to open transport to 208.49.103.92(208.49.103.92):9160 , java.net.ConnectException: Connection timed out
 INFO [Hector.me.prettyprint.cassandra.connection.CassandraHostRetryService-1] 2013-08-16 13:07:01,041 Not checking that XXX.XXX.XXX.XXX(XXX.XXX.XXX.XXX):9160 is a member of the ring since there are no live hosts

XXX.XXX.XXX.XXX = a public IP address. The agent can't hit the public IP.

I'm not getting any Cassandra metrics for that host in OpsCenter except for disk space.

What is the right combination of settings for my setup in address.yaml?

Thanks!