RWTH Compute Cluster

Get Access

1. Request a DKE cluster account here
2. You will recieve one email from Aachen and one from Maastricht
3. Following instruction in Aachen email to activate and create the cluster account
4. Reply to Maastricht email with newly created User ID and wait for reply
5. Go to your Aachen account here, select the Accounts and Passwords page, and set a new password for Hochleistungsrechnen RWTH Aachen and WLAN/VPN services

Login

1. Establish VPN connection to Aachen following instructions here
2. Login with your new RWTH Aachen account:

ssh <Your User ID>@login18-1.hpc.itc.rwth-aachen.de

Usage

• Check manual for commands here
• Here an example sbumit.sh job script:

#!/bin/bash

#SBATCH --job-name=cytoeffect_short
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=8
#SBATCH --mem-per-cpu=4GB
#SBATCH --time=01:00:00
#SBATCH --account=um_dke

R -e "rmarkdown::render('Reanalysis_Aghaeepour2017_Poisson.Rmd')"

• Submit the script to the scheduler:

sbatch submit.sh

• Check status (change YOUR_USERNAME to your username):

squeue -u YOUR_USERNAME

Example

First install RStudio. This will allow you to knit Rmd files.

RSTUDIO_VERSION=rstudio-server-rhel-1.2.1335-x86_64
mkdir $RSTUDIO_VERSION
cd $RSTUDIO_VERSION
wget https://download2.rstudio.org/server/centos6/x86_64/${RSTUDIO_VERSION}.rpm
rpm2cpio ${RSTUDIO_VERSION}.rpm | cpio -idmv
cd ..
echo "export RSTUDIO_VERSION=$RSTUDIO_VERSION" >> .zprofile
echo 'export PATH=$HOME/${RSTUDIO_VERSION}/usr/lib/rstudio-server/bin/pandoc:$PATH' >> .zprofile
source .zprofile

Now clone a repository and submit a job script. This example should take about 30 min. It will request one node with 8 cores for 1 hour.

git clone https://github.com/ChristofSeiler/blish_cytoeffect_tutorial.git
cd blish_cytoeffect_tutorial
sbatch submit.sh

Storage

• Backed up: $HOME
• Not backed up: $WORK and $HPCWORK