ejabberd - Comments for "Cluster does not start when running ejabberd"

cluster startup and documentation problem

tdewolf — Mon, 20 Jul 2009 07:52:10 +0000

documentation definitely needs improvements! I took me 3 working days to get a cluster up and running, which did me writing this blog.

Cluster Startup and Documentation Problem

dschaub — Thu, 17 Jul 2008 15:26:02 +0000

I have found a problem for the error in clustering.

Problem: Following the docs for creation of a cluster works up to the very last statement that says, "Start ejabberd normally".

Issue: When you are in the $BASE/bin directory and run the erl commands to create the mnesia database, the database is created as a subdirectory to bin, plus it has MNESIA on the front..... example $BASE/bin/MNESIA.ejabberd@machine2. When you start ejabberd normally, it is looking for the mnesia database in $BASE/database/ejabberd@machine2.

Solution: Before starting ejabberd on machine2 in the last step, copy the contents of the MNESIA directory under bin to a directory under the database.... But, leave the MNESIA. off the front of the directory name.

Bug - either the docs need to state this, or there is a flag when running the erl commands that tell mnesia where to put the data that should be entered. Either way, the docs need to be updated.

Anonymous wrote: and

mfoss — Wed, 19 Mar 2008 22:23:46 +0000

Anonymous wrote:

and _while_ i am in that shell
...
i just cant use node2 as an ejabberd server

Right, this temporary Erlang shell is only for you to configure replication of several Mnesia tables, as explained in steps 3 and 4.

After you manually configured some tables to be replicated, then you stop that Erlang shell, and start ejabberd as usual.

The problem is probably that the spool directory of the Mnesia table when you start the Erlang shell is not the same directory that is later read by ejabberd. A quick and dirty solution is to copy the spool files created by Mnesia to the final destination before starting ejabberd.

Same problem here. At least

Anonymous — Wed, 19 Mar 2008 13:15:59 +0000

Same problem here.
At least I am not the only one...would be glad to hear a fix.