Monitoring your jobs

Version 1.0

This first version is a Google gadget to monitor your jobs online if your cluster doesn't already have this feature.

If the cluster you are running your jobs on uses the SGE queuing system, you are familiar with the command qstat to check your jobs. The qstat command has an -xml option which produces an xml file with all jobs informations.

Although many clusters have a webpage to monitor jobs, it is usually full of informations you don't need. Thus the idea of a simple web app. to monitor simple informations from your jobs.

See it in action on this page.

Download

If you want to insert the code directly on your web page, click here.

You can also get the Google gadget by clicking on this button:

Snapshot

Installation

Just try it

From your cluster, create a xml file with the command:

qstat -xml > qstat.xml

Note that the web app. doesn't support the option -ext which gives many more informations. I want to keep it very simple.

Export the file qstat.xml to a web server and enter its url into the Gadget Preferences. That's it !

Set up the web app. nicely

You want the xml file to contain fresh datas to be display on the web.

You can create a script qstat2web.sh to produce the xml file and to update the file on the web. It may look like this:

#!/bin/csh

# Output files:

set file='~/www/xml/beagle.xml'

set file2='~/www/xml/beagle_all.xml'

# Get the queue info:

qstat -u $USER -xml > $file

qstat -xml > $file2

If your home directory under ~/www links to a webpage you are almost done. if not, just add an scp line to the script qstat2web.sh to send the $file and/or $file2 to a webserver.

Last trick is to add a crontab job which regularly executes the script qstat2web.sh in order to send fresh datas to the webserver.

It might looks like this:

*/10     *       *       *       * pathtoyourfile/qstat2web.sh

This will update your jobs info every 10 minutes.