0.9.13
   - Updated pbs_stat.py to handle the new gmetric (3.0.1) interface. All
     options must now be handled by /etc/gmond.conf instead of commandline
     options. If pbs_stat.py is invoked with -o/--old then it supports this
     interface, default is the NEW one!

0.9.11
   - Version 0.9.10 did contain the right patch for pbs_statjob() function.
     Sorry for the inconvenience. You still require pbs_python version 2.7.4 
     and higher.

0.9.10
   - Fixed an error in pbs_stat.py that is triggered by version 1.1.0p4 of
     torque (pbs_statjob fails). This was an bug in the pbs_python interface.
     pbs_python version 2.7.4 and higher solves this problem.

   - Updated the installation instructions.


0.9.9
   - Fixed an typo in job.php (2.5.4 version) Fixed by Carlos Falquez
   - Fixed an error for both web intefaces in job.php, Always fill in the
     processors value in queue page, Fixed by Carlos Falquez

0.9.8
   - Fixed an error in pbs_stat.py that is triggered by version 1.1.0p4 of
     torque (pbs_statjob fails). 

     Reported by: Keith Poirier

     You have only to update the client software.

0.9.7
  - Changed the 'Show only jobs for user:' behaviour. It will now only show
    jobs for the selected user and not the other user jobs.
    Suggested by: Davide Salomoni

    This is ported for both web interfaces. After this version support for web 
    interface 2.5.3 will be dropped.

  - Added support to make an debian client package with checkinstall.


0.9.6
  - Replaced an statement in pbs_stat.py. So it now also support python
    version 1.5 instead of python 2.0 and higher.
    Fixed by: Daniel Olson

  - Added an extra option to pbs_stat.py: 
  	-d|--domain <name>

    Default: Let pbs_stat.py decide which domain to use.
    Reported by: Davide Salomoni

  - Fix an error in assignments.php for web interface 2.5.4. Forgot
    to include clustername in an url.
    Reported by: Davide Salomoni

    This change requires that you have to update the client and web interface
    with version >= 2.5.4

0.9.2
  - Fix an error if openpbs/torque use long hostnames in nodes file.

    Thanks to: Fokke Dijkstra

0.9.1
  - Added a INSTALL note that the web interface for 2.5.5 is the
    same as for 2.5.4.

  - Some changes to the documentation. This interface works for
    OPENPBS and TORQUE

0.9
  - There are now two web interfaces:
  	- One for ganglia webfrontend 2.5.3 ( This was the old one )
	- One for ganglia webfrontend 2.5.4 ( The new one )

    Both are tested. I do not know if it works with earlier/later 
    versions of ganglia webfrontend.
	   
0.8.5
  - Reduced the info send by the gmetric command. The gmetric command can only
    handle FRAMESIZE long messages (mostly 1400 bytes). Jobs that uses a lot of 
    nodes crashed pbs_stat.py

    This change requires that you have to update the client and web interface!

    Reported by: Keith Poirier

0.8
  - Fixed a memory leak when running in daemon mode. This requires at least
    version 2.7 of the openpbs python module.

0.7.4
  - Fixed a bug in the interval code setting for daemon mode. Program
    crashed when using a other interval then 25 seconds.
  - We can also set the multicast channel. Patch supplied by Daniel
    Olson. 
  - Addded the multicast channel option to the init.d script
  - The directories were set group id (setgid) in the distribution.
    Fixed this. Thanks to Daniel Olson.

0.7
   - Added command line options: pbs_stat.py -h|--help (for list)
   - Added an example init.d script called: pbs_monitor.
   - Changed the syslog message to 'pbs_stat.py' instead of 'python'

0.5.1
   - Fixed a bug in pbs_stat.py. If queue is empty and running
     pbs_stat.py in cron mode then root would get a lot of emails
     with content: None

     Thanks to: Daniel Olson <dolson@aem.umn.edu>

0.5
   - Added a command line options to pbs_stat.py. It has two options
        cron)   pbs_stat.py cron
	daemon) pbs_stat.py daemon

   - Fixed a bug for PHP installations that are not configured with 
     register_globals on. Then the passing of variables goes wrong and
     all the URL's do not work. This type of configuration is default
     for PHP version 4.2 and higher.

     Thanks to: Daniel Olson <dolson@aem.umn.edu>
                Seth Graham <sether@fnal.gov>

   - Rewritten the code to determine how many processor a cluster has.
     Now it also works on SMP systems

   - Added queue attribute to the web interface and client software.
     So we can see in which queue a job is running or scheduled.

0.2

   - Socket.getfqdn() requires python 2.0 and higher
     replaced it by: socket.gethostbyaddr(socket.gethostname())[0]

     Thanks to: Daniel Olson <dolson@aem.umn.edu>
                Seth Graham <sether@fnal.gov>


   - gmetric send its output to eth1. Removed this and let gmetric
     decide which interface to use. Some left overs from the SARA
     setup

     Thanks to: Seth Graham <sether@fnal.gov>
