Introduction to the Cluster

• Open the terminal or PuTTY
• Type ssh yourusername@enigma2.jhsph.edu
  • Now you're on enigma!
• Type qrsh to get on a node
  • Now you're on a node! Do everything from a node except submitting batch jobs, which can only be done from enigma
• When you're done, type exit to go back to enigma
• Type exit again to exit enigma

• Change directory: cd
  • Subdirectory: cd [subdir]
  • Home directory: cd ~/
  • New directory: cd /new/path/
  • Go up a directory: cd ..
• Print current working directory: pwd
• List files: ls
• Create new directory: mkdir [dirname]
• Copy file: cp [filename]
• Delete file: rm [filename]
• Move or rename file:
  • Move: mv [filename] [newlocation]
  • Rename: mv [oldname] [newname]

• View or edit text files:
  • This includes .txt, .sh, .R, etc.
  • cat [filename] to view entire file
  • head [filename] or tail [filename] to preview file
  • less [filename] or more [filename] to scroll through file (hit q to exit)
  • emacs -nw [filename] to view and edit file
• Options with "-". Examples include:
  • ls -l lists files and includes file information
  • rm -f [filename] forces removal of file without prompting
  • matlab -nodisplay opens MATLAB on the command line (without the interactive display window)
• Tab completion (for file or directory names)
• Google is your friend!

1. Get onto enigma and qrsh to get onto a node
2. List the files in your home directory (that's where you are)
3. Create a new directory called "BST753" within your home directory
4. Copy a file from your home directory into BST753
5. Go into BST753 and look at the files
6. Go back to your home directory

• First, make sure you are on a node!
• Type R
• Do stuff!
• Reading and saving data
  • Read files from locations on the cluster
  • Save R data files (.Rdata, .rda) within your home directory
• Plotting
  • If X11 is set up, plots will display
  • Otherwise, save plots using pdf() or png() functions
• When finished, exit R by typing q()

1. Get on enigma
2. Go to the directory where your code is using cd path/to/code
3. Create a "shell script" that will run your R code using emacs -nw shellscriptname.sh
4. Write shell script (next slide)
5. Exit emacs and save file by typing Ctrl-X Ctrl-C then y
6. Submit job by typing qsub shellscriptname.sh

#!/bin/sh
#$ -cwd 
#$ -l mf=40G,h_vmem=40G,h_fsize=10G 
#$ -m e -M youremail@jhsph.edu

R CMD BATCH --no-save --no-restore Rfilename.R

• The Biostat IT group has several helpful pages:
  • Lecture on the cluster
  • Detailed page on using the cluster