Some useful things to know about upstart on smgl

- The jobs upstart knows about are in the *.conf files,
  one file per service/daemon/task/job. Any file not ending in .conf
  is ignored.

- The jobs MAY NOT FAIL, or the entire boot process is halted...
  Use a script in the scripts dir and a job  pointing to it when
  a job can fail and make sure the script exits cleanly.

- tty{2,3} start as soon as everything is mounted.
  Annoyingly, GDM (I'm not sure about the others) claims the 1st available
  tty once it starts, so if tty{2,3} didn't start at that point, they'll
  try to respawn forever, bothering GDM and ultimately bring down the CPU.
  So when reordering the scripts, keep in mind that the displaymanager script
  is started at least a second after tty{2,3}. tty1 is unaffected, as that's
  started as soon as your / is mounted.

- when writing a job, start on A and ( B or C ) won't work
  use a separate BorC job and use start on A an BorC in the original job.

feel free to post bugs with additional jobs on
http://bugs.sourcemage.org and assign them to abouter@sourcemage.org
I'll get to them eventually, or just put them in the smgl-upstart spell
yourself if you have git access ;)

A little info on how to use upstart, because there isn't that much available
unless you run it in sysvcompat mode...

starting and stopping jobs in upstart:

  start <jobname>

or

  stop <jobname>

where <jobname> is the jobs name in /etc/init without the .conf part.

for example, to start lighty:

  start lighttpd

for a list of jobs and their status:

  initctl list

to emit a signal, in this case to stop everything that depends on network
connectivity but not the network job itself:

  initctl emit --no-wait network-stop

the --no-wait means "don't wait around to see what is happening.. RUN!!"