cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From aaron morton <>
Subject Re: Dual NIC server problems
Date Tue, 14 Dec 2010 09:15:30 GMT
The code for nodetool appears to just pass the host value through to the NodeProbe. Was there
anything else in the stack trace ? 

If you use the host name of the machine rather than ip what happens? includes a link to this page about getting JMX running with firewalls 
Perhaps the info on  java.rmi.server.hostname may be helpful 

Hope that helps.

On 14 Dec 2010, at 21:36, Arjan van Ham wrote:

> I have a cluster with three nodes, version 0.7.0 RC2. Each node has dual NIC's, eth0
to the internet and eth1 to a private network ( The outside NIC  on each node
is frewalled using iptables, only port 22 is allowed through. My cassandra.yaml configuration
file refers only to the internal hostnames that resolve to addresses.
> Most of my Cassandra installation works fine, except for a few nodetool commands. It
seems that they are trying to access the servers over the outside (firewalled) NIC and thus
fail. If I disable the firewall they work fine. 
> For example:
> [cassandra@node11 ~]$ nodetool -h decommission
> Error connecting to remote JMX agent!
> java.rmi.ConnectException: Connection refused to host:; nested exception
> Connection timed out
>        at sun.rmi.transport.tcp.TCPEndpoint.newSocket(
>        at sun.rmi.transport.tcp.TCPChannel.createConnection(
>        at sun.rmi.transport.tcp.TCPChannel.newConnection(
>        at sun.rmi.server.UnicastRef.invoke(
>        at Source)
>        at
>        at
>        at
>        at
>        at<init>(
> And if I disable the firewall on the remote node it works right away.
> Any ideas anyone?
> Thanks.

View raw message