Cluster Specific Documentation

CCBB actively used cluster is Phylocluster. For overall cluster user's benefit please read this Phylocluster Access and Responsible Use Policies.

CCBB clusters are based on the ROCKS cluster software. They currently use the Sun Grid Engine batch queuing system to manage jobs. In order that the nodes of the cluster not be overloaded, you will be expected to submit jobs to SGE which then sends the jobs out to the nodes of the cluster for execution. In order not overload the nodes, SGE will schedule your job. This relieves you of the responsibility of having to do this at cost of having to learn how to edit job script files, and also how to submit, monitor, and manage jobs.

We have some resources for this. First, we have a tutorial showing various aspects of the using the SGE system. Also, more advanced users may want to use task arrays to process multiple job scripts. SGE also supports several multiprocessing paradigms. Even if you are not a programmer, you may want to review this page so you will be familiar with the technologies. Then if you see that a program you do want to use supports a given technology, you'll be able to use that program correctly (because, for example, submitting a job which uses 8 processors when you only tell SGE it will use 1 can cause SGE to make bad decisions with regards to scheduling), and to yield the maximum amount of performance gain.

Note that the ROCKS clustering software is the CentOS Linux Operating System with some extensions for managing the cluster nodes, the queuing software. To effectively use the cluster, YOU MUST KNOW UNIX!!!! Please refer to our UNIX page for more information on resources to make this easier.

Please see our Software page for information on what software we have installed.