Getting started with the cluster

The cluster head node is Use ssh from a terminal, or with an application such as PuTTY, to log in. If you are on a computer that is part of FASnet, the head node is known as Please be sure to follow the instructions below the red button on the documentation page before using the cluster for the first time. In particular, note that you must join the mailing list called colony-users before you can use the cluster.

The cluster can be used in interactive or batch mode.

Interactive mode

Log on to queen and start an interactive job with the qsub command. For example:

qsub -I -V -l walltime=01:00:00,nodes=1:ppn=1,pmem=1gb -W group_list=colony-users

will start an interactive job (-I) of 1 hour on one processor of one node having at least 1GB of memory.  Once logged on to a node of the cluster, you can use it to do your computing from the command line, or you can start a ``virtual desktop''. Command line is good for running scripts that do not require graphics. The virtual desktop is necessary if you want to start graphics windows.

Command line

Software is available through ``modules'', which must be loaded on login. For example,

module load LANG/R/3.0.2

will make R version 3.0.2 available to you. You can now type R at the command line to start R. For a list of all available modules, type

module avail

For more information on modules, see the FASnet documentation.

Virtual desktop

A virtual desktop is a window on the computer you are working from (e.g., your laptop) that shows a desktop environment on the cluster node. On this virtual desktop you can run R, RStudio, a web browser, acrobat, etc. The software to enable the virtual desktop is called vnc. You will need to

  1. Start a vnc server on the cluster node you have logged into
  2. Start a vnc client on your computer.

To use RStudio start a terminal from the Applications -> System Tools menu, and load the appropriate module with:

module load TOOLS/RSTUDIO/0.98.944

This will also load the module for the most up-to-date version of R. Now type rstudio in the terminal window to start RStudio.

Batch mode

Below are three templates for computing with the colony cluster that may be useful for getting started. This is not a replacement for the HPC wiki which contains complete documentation. To run jobs on the cluster you need a portable batch script (PBS) file that is submitted from the head node of the cluster, The PBS file specifies resources required to run the job and the command to execute. In the second and third examples below, the command to run is R with an input script.

  1. The first template is an example of running a hypothetical program called `myprogram' several times with different input files as an "array job". See the PBS file.
  2. The second template is an example of running a simulation study in R with several different simulation configurations using the R package parallel. See the PBS file and the R script.
  3. The third template is an example of distributing a loop across nodes of the cluster using the R package parallel. See the PBS file and the R script.