[ejabberd] Setting up cluster at Amazon environment

tyju tiui jckdnk111 at yahoo.com
Tue Feb 17 00:39:50 MSK 2009


Hi Max,

I'm running a cluster on EC2 with no problems.
You might want to check your ejabberdctl.cfg ... the last line should be something like "ERLANG_NODE=ejabberd at hostname"
I had to edit /etc/hosts on all cluster members to map the hostnames of all cluster member private IP's (EC2 internal IP's).

So lets say you have (2) cluster members: 
1) hostname "chat1" with ip=10.254.111.79 (this is domU-12-31-39-00-68-A1.compute-1.internal)
2) hostname "chat2" with ip=10.254.138.212 (this is domU-12-31-39-00-85-26.compute-1.internal)

in chat1's /etc/hosts file add: "10.254.138.212 chat2"
in chat2's /etc/hosts file add: "10.254.111.79 chat1"

Also, make sure you have the following in your ejabberdctl.cfg on chat1: "ERLANG_NODE=ejabberd at chat1"
and the following in your ejabberdctl.cfg on chat2: "ERLANG_NODE=ejabberd at chat2"

Goddluck!

Ty




________________________________
From: Maxim <sevmax at gmail.com>
To: ejabberd at jabber.ru
Sent: Monday, February 16, 2009 4:12:49 PM
Subject: [ejabberd] Setting up cluster at Amazon environment

Hello, community.
I have run 2-node cluster in Amazon Web Service environment. This is Computing Cloud, which give access to virtual servers hosted on XEN. Each server have unique internal DNS name, like these:
node1 is node1 at domU-12-31-39-00-68-A1.compute-1.internal
node2 is node2 at domU-12-31-39-00-85-26.compute-1.internal

both of nodes have external IPs, which configured for site: chat.example.com
DNS provide give us roundrobin balancing, but all xmpp client are connected to first node (as its IP less then for second one). There is why I need in interconnection of nodes and balancing by ejabberd tools.

When I try to make interconnection I receive next error:

#erl -sname node1 -remsh node2 at domU-12-31-39-00-85-26.compute-1.internal

Erlang (BEAM) emulator version 5.6.3 [source] [async-threads:0] [hipe] [kernel-poll:false]
{error_logger,{{2009,2,16},{15,1,59}},"** System NOT running to use fully qualified hostnames **~n** 
Hostname ~s is illegal **~n",["domU-12-31-39-00-85-26.compute-1.internal"]}
*** ERROR: Shell process terminated! (^G to start new job) ***
=ERROR REPORT==== 16-Feb-2009::15:01:59 ===
** System NOT running to use fully qualified hostnames **
** Hostname domU-12-31-39-00-85-26.compute-1.internal is illegal **

How can I run interconnection for cluster nodes?
I think the reason of failed is in naming of instances.
What do community think about this situation?

Regards,
Max 


      
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.jabber.ru/pipermail/ejabberd/attachments/20090216/c356a1b4/attachment.htm>


More information about the ejabberd mailing list