I believe Tony Vroon <chainsaw@gentoo.org> has been handling ejabberd stuff... This may be some strange error on my part as I was running ejabberd for a while with no problems. I can't remember if this broke after the upgrade to 1.1.1 or not. When I run /etc/init.d/ejabberd start I receive a failure ([ !! ]), but the server is actually started and seems to be running fine. I can see it running through ps with a line including the following: /usr/lib/erlang/erts-5.4.6/bin/beam -- -root /usr/lib/erlang Could it be that start-stop-daemon searches for a process with the name ejabberd since that's what it tries to start? I've heard that start-stop-daemon has issues with wrapper scripts.
Jep! I see exact this Error too ..
I suggest that ejabberd follows the lead of how debian does it: using su instead of start-stop-daemon. After all, if we're not using start-stop-daemon to stop perhaps it shouldn't be used to start... I'll attach a debian-inspired init.d script for review. Since it uses su it requires a real shell for the jabber user... is that ok?
Created attachment 95382 [details] files/ejabberd-1.1.1.initd
Created attachment 95383 [details] Debian init.d file Attaching Debian's init.d file for comparison
Hello! The new init.d script is not working here :( BlackBox init.d # /etc/init.d/ejabberd start * Starting eJabberd ... This account is currently not available. [ !! ]
Ok, sry! Its my fault ;) Forget to set a valid shell! Now, the init.d script works fine. start, stop and reload!
Jakub: net-im/ejabberd != net-im/jabberd
Add folowing lines to init.d script. After line "start-stop-daemon ....." and before line "eend $?" insert lines: if [ -z "$EJABBERD_NODE" ]; then EJABBERD_NODE="ejabberd@`hostname -s`" fi /usr/bin/ejabberdctl $EJABBERD_NODE status | grep 'Status: started' Problem solved on my system with ejabberd-1.1.1.ebuild for amd64 arch. ps Why 1.1.1.ebuild for amd64 not exists?
Created attachment 98875 [details] baselayout-compatible init.d I've attached a new init.d script that works for me with baselayout-1.12.5 on amd64.
Incidentally, when this change is accepted into the tree, #141302 should be pinged to keyword this ~amd64. The software itself is terrible, especially its logging, but the ebuild is as good on amd64 as on x86.
+*ejabberd-1.1.1-r1 (12 Oct 2006) + + 12 Oct 2006; Tony Vroon <chainsaw@gentoo.org> + +files/ejabberd-1.1.1-r1.initd, +files/self-cert-v2.sh, +files/ssl.cnf, + ejabberd-1.1.1.ebuild, +ejabberd-1.1.1-r1.ebuild: + Revision bump, robust start/stop script thanks to Chris Carlin + <ccarlin@physics.tamu.edu> and Dustin J. Mitchell <dustin@v.igoro.us> in bug + #145373. Statsdx & SOCKS5 proxy support thanks to Nikolaus Polak + <nik@linuxlovers.at> in bug #137724. And finally, thanks to Micha Krause + <linux@krausam.de> for pointing out that the SSL eclass did not create + certificates usable for S2S connections in bug #150088. This should fix it then. My apologies for how long this bug has lingered, it was not assigned to the correct person.