IDE Setup

With access to the PMACS server (see PMACS Usage), here are some IDE options:

VS Code
Jupyter Notebook

The Posix server supports both, but caps you at 20GB memory — these approaches let you request more.

Setting Up VS Code for PMACS HPC

Two approaches: Interactive Session (simple, capped at 8 hours) or Alternate Method (scripted, longer sessions).

Step 1: Install VS Code Locally

Download from the official page, sign in with GitHub or Microsoft (required for tunnel auth), and install the Remote - Tunnels extension.

Step 2: Install the VS Code CLI on PMACS HPC

(Already done as of 4/15/25 — pinned to v1.85 because the server’s GLIBC 2.17.0 isn’t supported by v1.86+.)

wget -O /project/hipaa_ycheng11lab/software/vscode_1.85/vscode-server.tar.gz "https://update.code.visualstudio.com/1.85.2/cli-alpine-x64/stable"
tar -xzf /project/hipaa_ycheng11lab/software/vscode_1.85/vscode-server.tar.gz -C /project/hipaa_ycheng11lab/software/vscode_1.85

Add to PATH if needed:

echo $PATH | grep --color "/project/hipaa_ycheng11lab/software/vscode_1.85"
echo '# Include the VS Code Server executable from the shared installation in PATH' >> ~/.bashrc
echo 'export PATH="/project/hipaa_ycheng11lab/software/vscode_1.85:$PATH"' >> ~/.bashrc
source ~/.bashrc
which code   # should show .../vscode_1.85/bin/code

Approach A: Interactive Session (up to 8 hours)

bsub -n 4 -R "rusage[mem=8000]" -Is bash
code tunnel

Authenticate if prompted. In local VS Code, open Remote Explorer → Tunnels, find your session by name and connect (or use Command Palette → Remote-Tunnels: Connect to Remote Tunnel).

Approach B: Scripted (longer sessions)

Pre-authenticate:

code tunnel user login --provider github
code tunnel user show

Submission script (vscode_server.sh, example requests 48 hours):

#!/bin/bash
#BSUB -J  vscode_server
#BSUB -W 48:00
#BSUB -n 4
#BSUB -M 51200
#BSUB -o vscode.%J.log
#BSUB -e vscode.%J.error

code tunnel --accept-server-license-terms

bsub < vscode_server.sh
bpeek <JobID>   # view output; if not pre-authed, follow the device-login URL shown here

Then connect from local VS Code the same way as Approach A.

Notes: SSH into worker nodes is restricted and doesn’t persist across node changes — tunnels follow you instead. Interactive queue caps at ~8 hours (bqueues -l interactive). Stuck tunnel → reload VS Code window or rerun the script.

Setting Up Jupyter Notebook for PMACS HPC

Mac

Instructions borrow from the HPC wiki, updated as of 02/19/2025. Windows/MobaXTerm follows the same steps with a GUI for port forwarding at step 3.

Log into an interactive session, cd to your working directory.

 bsub -Is bash
 module load python/3.11 gcc/12.2.0   # gcc/10.2.0 also works
 pip install -U jupyter_nbextensions_configurator jupyter_server_proxy scikit-learn   # first time only

Start Jupyter:
```
 jupyter notebook --ip $(hostname)
```
Note the node name and the http://127.0.0.1:PORT/?token=... URL in the output. Leave this running.
In a new terminal on your Mac/Linux machine, tunnel in (match the node name and port from step 2):
```
 ssh -L 8888:node157:8888 <your_username>@consign.pmacs.upenn.edu
```
Paste the http://127.0.0.1:... URL from step 2 into your browser. (Unique per session — not bookmarkable.)

Windows MobaXTerm

Same result, replaces step 3 above:

Tools → MobaSSHTunnel (port forwarding) → new SSH tunnel (start it only after the Jupyter server is running).
Fill in: MyComputer port 8888; SSH Server PMACS_USERNAME@consign, port 22; Remote Server nodeNODENUM.hpc.local, port 8888 (unless the URL says otherwise). Save, name it, start it.
Paste the URL into a browser — you should see the Jupyter GUI, and the CLI should report a GET request.