[ejabberd] freebsd upgrade and now auth failures

Randy Bush randy at psg.com
Fri Apr 10 16:32:13 MSD 2009


thanks to you and badlop for help.  much appreciated.

>> work0.psg.com:/root# ejabberdctl mnesia
>> RPC failed on the node 'ejabberd at jabber.psg.com': nodedown
>> 
>> =ERROR REPORT==== 9-Apr-2009::13:28:41 ===
>> Error in process <0.33.0> on node 'ejabberdctl at work0.psg.com' with exit value: {badarg,[{erlang,list_to_existing_atom,["ejabberd at localhost"]},{dist_util,recv_challenge,1},{dist_util,handshake_we_started,1}]}
> 
> ejabberdctl just connects to your ejabberd via a normal erlang remote 
> shell connect. the message is erlang telling you, that the name of the 
> process it tries to connect to is not the one you have given in your 
> connection attempt via ejabberdctl (see the script for how it does
> this).

ok, i have this sorted.  some of the problem was that, as you said,
there are hidden dependencies on the start-up script.

i can now run as any one of
    ejabberd at localhost
    ejabberd at jabber.psg.com  (the of service and a secondary dns name
    ejabberd at work0.psg.com   (the true host name)

so, for each of the above, i
    o kill ejabberd, beam, ...
    o edit all control files to choose one of the above names
    o restore /var/spool/ejabberd from before the upgrade
    o reboot and let the server restart ejabberd normally 
    o ejabberd starts successfully, and
    o ejabberdctl status shows it running

in each case, ejabberd comes up.  but a
    ejabberdctl dump /tmp/foux
does not have the old users and buddy lists.

this is getting pretty desperate.

randy


More information about the ejabberd mailing list