Ejabberd 2.0 Clustering Problem

Instead of

"[’ejabberd@dev-im1’]"

you should use

"['ejabberd@dev-im1']"

This is to start an erlang node with ejabberd running in it. Among other things, ejabberd will start mnesia.

And this is to start an erlang node with only mnesia running, and telling mnesia that there is another node.

Until here, all looks correct.

Oh, im1 should be listed in the 'running db nodes' list.

You can try a very simple setup, without using ejabberd at all, to verify the basic concepts. Maybe it gives you some idea of where your problem is.

For example, you can start two simple erlang nodes (don't care about ejabberd right now), in the same machine.

Console 1:

$ erl -sname serv1@localhost -s mnesia
Erlang (BEAM) emulator version 5.6.2 [source] [smp:2] [async-threads:0] [kernel-poll:false]

Eshell V5.6.2  (abort with ^G)
(serv1@localhost)1> mnesia:system_info(running_db_nodes).
[serv1@localhost]

Console 2:

$ erl -sname serv2@localhost  -mnesia extra_db_nodes "['serv1@localhost']" -s mnesia
Erlang (BEAM) emulator version 5.6.2 [source] [smp:2] [async-threads:0] [kernel-poll:false]

Eshell V5.6.2  (abort with ^G)
(serv2@localhost)1> mnesia:system_info(running_db_nodes).
[serv1@localhost,serv2@localhost]

Now again in console 1, to verify the first node also noticed the second is connected to it:

(serv1@localhost)2> mnesia:system_info(running_db_nodes).
[serv2@localhost,serv1@localhost]

Resolved:

The Manual says that only port 4363 needs to be open to communicate with other erlang nodes. However, erlang uses EPMD for communication between nodes and it opens up random ports for communication. As soon as i opened up all ports between only the two nodes i was able to interconnect them. I think this should be added as either a footnote or put into a troubleshooting section for clustering. Also i was wondering if anyone knew how to open up only specific ports for the epmd? I don't like the idea of opening up all ports even thought it is only between these machines.

Thanks
Andy