Command Line

Client Access

Setup options are on the HPC wiki. Mac’s Terminal app works but is missing some commands (see Personalization below). Install a free FTP client (WinSCP, CyberDuck, Filezilla) for a GUI file transfer — use the SFTP connection type.

Logging In

First time: follow the HPC wiki’s First Login steps. Otherwise:

$ ssh <PMACS_ID>@consign.pmacs.upenn.edu

If the password prompt never comes and the connection times out, check you’re on the VPN.

Command Line Personalization

Your .bashrc runs on every new terminal/job submission. To use lab-installed software, add to it:

if [ $HOSTNAME != "consign.hpc.local" ] &&
   [ $HOSTNAME != "mercury.pmacs.upenn.edu" ] &&
   [ $HOSTNAME != "hpclogin.pmacs.upenn.edu" ] &&
   [ $HOSTNAME != "hpclogin1" ]; then
    LAB_SOFTWARE="/project/hipaa_ycheng11lab/software/"
    SOFTBIN="${LAB_SOFTWARE}/bin/"
    PATH=$SOFTBIN:$PATH
    export PATH
fi

On your own machine, Mac users should install Homebrew (apt-get/yum/pip’s equivalent):

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

then e.g. brew install wget.

Windows/MobaXTerm: Settings → Configurations → Terminal Features → set logging to printable output with timestamps (very useful for later debugging). You can also create SSH sessions for Mercury/Consign — right-click the session → Edit Session → Advanced SSH Settings → Execute Command newgrp hipaa_ycheng11lab, so files you create are shared with the lab.

Interactive Nodes

Mirrors the HPC User Guide — worth a read for more detail.

Use an interactive node for compute-intensive jobs rather than running directly on the head node (consign.pmacs.upenn.edu):

$ bsub -Is bash
$ bsub -n 4 -R "rusage[mem=75000] span[hosts=1]" -M 75000 -Is bash   # with more memory/cores

For a Jupyter notebook within the node (Mac steps — see Tutorials → IDE Setup for the full walkthrough; other OSes: HPC wiki):

cd to your working directory first (Jupyter’s GUI navigation between subdirectories is limited).
On the interactive node: jupyter notebook --ip $(hostname)
In another local terminal, tunnel in: ssh -L 8888:node###:8888 <your_username>@consign.pmacs.upenn.edu (match node name/port to the output)
Open the http://127.0.0.1... link from the notebook output — unique per session.

Submitting Jobs

Basic:

$ bsub <script_name>

With parameters:

$ bsub -J IHRA_markers -e IHRA_markers.%J.error -n 8 -R -M 1024 'rusage[mem=1024] span[hosts=1]' bash IHRA_gene_marker_predicates.bash

Flag	Meaning
`-J IHRA_markers`	LSF job name
`-o IHRA_markers.%J.out`	output file (needed to receive one)
`-e IHRA_markers.%J.error`	error file (needed to receive one)
`-n 8`	cores requested
`-M 1024`	MB of memory
`-R 'rusage[mem=1024] span[hosts=1]'`	ensure enough memory + all cores on one node
`bash`	job type
`IHRA_gene_marker_predicates.bash`	script name

Or run an LSF job script directly (parameters live in the file’s header):

$ bsub < <script_name>

#!/bin/bash
#BSUB -J JNAME
#BSUB -o JNAME.%J.out
#BSUB -e JNAME.%J.error
#BSUB -n 1
#BSUB -M 10
#BSUB -R "rusage[mem=10] span[hosts=1]"
#BSUB -notify done         # email when the job finishes
#BSUB -u <username>@pennmedicine.upenn.edu   # required alongside -notify

echo "hello"