The following files are available for download. Please note that if you wish to download Clustor, you can obtain it from the ActiveTools site.

All the nodes in our cluster are dual processor. So we allow each user a maximum of two jobs per node while each node can have a total of five jobs run (by all users). This configuration can be seen in the clustor.options file. We use a scheme whereby long running jobs are given lower priority. A cron job is regularly run which checks how long each job has been running and if it has run for more then a certain time it reduces it's priority accordingly. These current job lengths and associated nice levels are shown in Figure 1 shown at the right while the ReNice script can be downloaded.

Length (min) Nice Level 0 - 30 0 30 - 60 5 60 -120 10 >120 20 Figure 1

To aid in determining the status of the nodes and jobs running on them we have written several scripts. Some of these are accessible from the command line and some via a web interface. The script clientconfig lists the configuration of all the nodes, checking whether they are booted into Linux or Windows NT. This information is also available on the web via this script. The script nodeload checks who is running clustor jobs on each node, and how many are running.