* Philipp Huebner <debala...@debian.org> [2014-08-19 10:41]: > Could you please clarify whether this is internal behaviour of ejabberd > or related to handling the configuration files? > > For example, was EJABBERD_NODE previously set (manually / automatically) > in /etc/default/ejabberd or not?
The wheezy package sets EJABBERD_NODE=ejabberd both in ejabberdctl(8) and in the Debian-specific ejabberd(8) script. Erlang auto-expands this node name to ejabberd@$(hostname). Your experimental package uses upstream's default instead: EJABBERD_NODE=ejabberd@localhost IMO, this is a saner default, as it saves you from trouble[*] when ejabberd is moved to a new box, or the local host name is changed, or the host name doesn't resolve to a local IP address for some reason. The only occasion where you'd have to use a non-localhost node name is when you set up clustering, but in that case, you have to tinker with the configuration (and know what you're doing) anyway. That said, just setting EJABBERD_NODE=ejabberd as the package did in the past is probably the easiest way out of upgrade issues. If you don't handle this issue one way or another, each and every upgrade will fail. [*] http://www.process-one.net/docs/ejabberd/guide_en.html#changeerlangnodename -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org