Improvements:

* Auto restart spoolwatcher's watch

  * use dbus to detect newly mounted drive (hard to do)
  * spoolwatcher hangs out on jabber and accepts a "make new drive" command
    (which runs a script to format/mount/watch the drive) (security hole)
  * a script is added to jumpgate which does the drive formatting/mounting/etc
    and then when finishes sends an xml-rpc message to spoolwatcher to
    start watching again. 

* Change umask group for the rsync to be writable for multiple users
  really really needs to be 002
* the directory also needs to be set group id
* need to make sure that there's a final rsync that finishes without coping any files
* demon processes need to restart

  * most likely solution, they should detect if there's a currently running 
    version and stop running. This'd allow a cron script to restart them

* The machine to machine communication needs to be via xml-rpc

  * Update benderjab and/or xmppy to send/receive xml-rpc messages

* Add logging to everything

  * high priority messages go out as jabber messages
  * low priority go to disk
  * some subset of recent messages should be stored in ram so they can 
    be retrieved by a user jabber status message (or a web page view)

* Generate config files

  * For goat (from brandon's web interface)
  * For bed file generator

* Need longish term storage to make generating bed and mapped read files
  from multiple lanes on multiple flow cells

* View current status page.

  * once xml-rpc is working it should be easier to update the jumpgate
    web page to report the current status of a run.

    * sequencing
    * waiting on copy (final rsync)
    * running goat
    * running pipeline
    * ... (more processing)