[ejabberd] Ejabberd errors on second node

Evgeny Khramtsov xramtsov at gmail.com
Mon Jun 9 14:19:52 MSK 2014


Mon, 09 Jun 2014 13:23:53 +0400
Regna <ki at bakka.su> wrote:

> Hello all.
> 
> I have 2 nodes:
> 1: ejabberd at cecilia (cecilia.int.bakka.su)
> 2: ejabberd at iris (iris.int.bakka.su)
> Both running ejabberd-2.1.13 with odbc mysql from same packages,
> mariadb-10.0.10 (master-master multi-source replicated), same configs
> with only difference being ip addrs and node name in ejabberdctl.cfg.
> 
> On 2'nd node, when I try to start ejabberd, I get this error:
> http://paste.bakka.su/v/N0
> Endless loop of those.
> 
> configs: ejabberd.cfg: http://paste.bakka.su/v/O
> 	 ejabberdctl.cfg: http://paste.bakka.su/v/E
> 
> mnesia:info(): http://paste.bakka.su/v/6I
> 	       http://paste.bakka.su/v/iu

Well, yes, this is a known problem: RDBMs startup is poorly written
which is why sometimes it leads to this kind of problems. In your case
sql_pool somehow is disc_only table on iris. This is wrong. Change it
from `ejabberdctl debug`:
> mnesia:add_table_copy(sql_pool, 'ejabberd at iris', ram_copies).


More information about the ejabberd mailing list