[13:36] <praveen_pk> When I try to start /sbin/init in a chroot of RHEL6, it complains about not being able to connect to /com/ubuntu/Upstart, Does init need d-Bus at Startup?
[17:57] <csgeek> is there a user's mailing list to subscribe to?  for asking questions.. I saw a devel list.. but I presume that's for new features being added to upstart itself. 
[17:59] <wraiden> what is your question?
[18:09] <csgeek> well.. I have an upstart script that starts up fine, but it fails to terminate properly 
[18:09] <csgeek> one sec.. I'll do a pastie 
[18:12] <csgeek> wraiden:   http://pastie.org/2238507  
[18:22] <wraiden> mhm
[18:23] <wraiden> the supervision of the process works?
[18:26] <wraiden> do you terminate with "initctl stop jobname"?
[18:26] <csgeek> I used sudo stop jobname 
[18:27] <wraiden> thats a shortcut...
[18:27] <wraiden> ok
[18:27] <wraiden> "initctrl status jobname" lists the right pid ?
[18:29] <wraiden> it has to be the right pid as it is the pid of the process that gets a sigterm to be stopped
[18:31] <csgeek> hmm.. the PID it captures seems off 
[18:34] <wraiden> there is your problem
[18:36] <wraiden> you have two exec in your main process...
[18:36] <wraiden> the "expect fork" follows only the first
[18:36] <wraiden> you can try "expect daemon"
[18:40] <wraiden> mhm
[18:40] <csgeek> okay.. does that the job still need to be backgrounded?  ie.  scriptname &  ?  
[18:40] <wraiden> i think so
[18:40] <wraiden> the exec stuff that i talked was rubbish...
[18:41] <wraiden> the execs use the same process environment so there are no forks
[18:41] <wraiden> try without expect fork
[18:44] <csgeek> okay.. one sec.. rebooting to reset state  and all that (glad i'm doing this on a vm ) 
[18:48] <csgeek> okay.. seems to work now
[18:49] <csgeek> ended up removing fork/daemon statements
[18:49] <csgeek> and i'm not backgrounding the job 
[18:55] <danolj> question regarding redis upstart script for ubuntu, more about redis than the upstart script
[18:55] <danolj> here is the proposed upstart script http://pastebin.com/n1nHhnF6
[18:55] <danolj> actually, this is an upstart question, sorry
[18:56] <danolj> the line 'pid file $PID_FILE' 
[18:56] <danolj> as I've read in the docs indicates to upstart
[18:56] <danolj> that it should monitor this process by the pid file, correct?
[19:00] <wraiden> what version of upstart?
[19:00] <danolj> 0.6.5-8
[19:01] <wraiden> man 5 init states "pid file" ?
[19:02] <csgeek> I'm on 11.04 natty.  so..     0.9.7
[19:02] <wraiden> i think this was before the supervision was changed to use signals from the kernel to the init process...
[19:02] <wraiden> i use upstream 1.3
[19:03] <wraiden> not on ubuntu at all *g*
[19:03] <wraiden> a quick grep over souces went negative
[19:03] <wraiden> *sources
[19:04] <wraiden> there is no pid stanza in 1.3
[19:06] <danolj> ok, thanks
[19:06] <danolj> so, perhaps I should simply have redis not daemonize itself so that init/upstart can retain a handle on it, sound good?
[19:08] <csgeek> okay.. I have one last question.. .. I wanted to set it up so when I call childJob, it knows that its requires parentJob, so if parentJob isn't running it'll start it then, start the childJob?  any wiki/doc page I can look at that talks about the process?  
[19:15] <wraiden> danolj: that would be the first thing to use if a service provides a "don't fork or daemonize"-option ...
[19:16] <danolj> redis provides the choice
[19:16] <wraiden> read the http://upstart.ubuntu.com/cookbook/
[19:16] <danolj> to confirm, don't daemonize redis if using init/upstart as the supervisory process
[19:16] <danolj> wraiden: okay, thanks for the pointer!
[19:17] <danolj> oh, yes, I have been reading that
[19:18] <danolj> found section 6.8.2 regarding daemonizing, thanks
[19:26] <wraiden> csgeek: you can hook in on starting and stopping events of the child with the parent but if parent should be running without child by default but could be stopped it could be scripted in pre-start with help of initctl
[19:28] <csgeek> thanks for all the help wraiden
[19:30] <wraiden> you're welcome...
[19:32] <wraiden> a port of the rhel cluster suite init scripts is a hell of an effort...