logo

Epigenomic Data Analysis 2016

Using the Guillimin HPC


Using the Guillimin HPC

This document provides information on how to get started with using the Guillimin HPC.

Connecting to the Guillimin HPC scheduler:

Host: guillimin.clumeq.ca

Login / Password: Provided to you on a separate piece of paper at the beginning of the workshop.

Note: Your home folder is limited to 10GB of storage. The tutorials we will do shouldn’t make you go over this quota, but to be aware of the remaining space you have, you can use the command myquota.

Linux/Mac

Use the following command from the command line:

ssh username@guillimin.clumeq.ca

The username and password are the ones that we provided on a separate piece of paper.

Windows

Connect to the guillimin.clumeq.ca server using the Putty tool that you installed for this workshop.

The first time you log in

You should set environment variables to ensure that the CVMFS modules will always be accessible to you and to the jobs you will launch. To do so, we’ll add a few command to the ~/.bashrc file.

[class99@lg-1r17-n04 ~]$ echo 'export MUGQIC_INSTALL_HOME=/cvmfs/soft.mugqic/CentOS6' >> ~/.bashrc
[class99@lg-1r17-n04 ~]$ echo 'module use /cvmfs/soft.mugqic/CentOS6/modulefiles' >> ~/.bashrc
[class99@lg-1r17-n04 ~]$ source ~/.bashrc
  • The $MUGQIC_INSTALL_HOME environment variable is required by some CVMFS modules.
  • The module use /cvmfs/soft.mugqic/CentOS6/modulefiles command will load CVMFS modules in your list of available modules.

Loading a module

All available software are organized as modules that you can load. For example, to start using Java 7, you need to load the Java 7 module.

1- Get the list of available modules

[class99@lg-1r17-n04 ~]$ module avail

2- Load your module of choice

[class99@lg-1r17-n04 ~]$ module load mugqic/java/openjdk-jdk1.7.0_60

3- You can now run the module you loaded directly

[class99@lg-1r17-n04 ~]$ javac -version
javac 1.7.0_60-ea

Launching a job on the scheduler

Note: Complete documentation on the Guillimin scheduler is available here: http://www.hpc.mcgill.ca/index.php/starthere/81-doc-pages/91-guillimin-job-submit

Jobs are submitted to the scheduler using the qsub command. The important qsub parameters for this workshop are the following:

  • nodes : total number of machines to use for this job
  • ppn : total number of cores to use for this job
  • -d workdir : The working directory, in other words the path in which this command will be executed

Please keep in mind that you need always need to load the modules you want to use in your scripts.

Using a bash script

You can specify a bash script to launch as a compute job directly from the command line, like the testjob.sh script in this example:

qsub -l nodes=1:ppn=1 -A bem-651-ae testjob.sh

where the content of testjob.sh could be:

#!/bin/bash
# PBS -l nodes=1:ppn=2

module use /cvmfs/soft.mugqic/CentOS6/modulefiles/
module load mugqic/bowtie2/2.2.4
perl bismark_genome_preparation --bowtie2 --verbose .

Specifying the whole command from the prompt

If you do not want to put your command in a script, you can write it directly in the prompt using the echo command like this:

    echo "touch test.txt" | qsub -l nodes=1:ppn=1 -d .

Viewing the current state of your jobs

Use the following command to see the current state of the jobs you launched:

showq -uclass99 -n -v
  • Idle means the job is waiting for its turn in the queue
  • Running means the job is currently running on one of the execution nodes

It can sometimes take up to a minute for your jobs to appear in the showq output.

Downloading a file from Guillimin to your computer

Linux/Mac

Use the following from the command line:

cd /path/to/destination/folder
scp user99@guillimin.clumeq.ca:/home/user99/myfile.txt .

To download a full directory, add the -R option

cd /path/to/destination/folder
scp -r user99@guillimin.clumeq.ca:/home/user99/myfile.txt .

Windows

Use the WinSCP software that you installed for this workshop, connecting to guillimin.clumeq.ca.

View on GitHub