A helper tool for remote debugging on HPC clusters.
Project description
🚀 remote-debug
A CLI tool to simplify visual debugging of Python scripts on remote HPC clusters directly from your local VS Code instance.
remote-debug helps you bridge the gap between your local editor and a script running on a remote compute node, making it easy to debug GPU-specific issues or complex cluster jobs with a full-featured debugger.
Table of Contents
Installation
Install from PyPI:
pip install remote-debug
Or, build from source using Pixi:
# Install pixi with: curl -fsSL https://pixi.sh/install.sh | sh
pixi install
Quick Start
-
Initialize your project This command creates the necessary VS Code launch configurations in
.vscode/launch.json.rdg init
[!WARNING] If
launch.jsonexists but is malformed, it will be backed up tolaunch.json.bakand a new file will be created.
-
Run your script on the cluster Prefix your usual Python command with
rdg debug. This will start a debug server and wait for you to connect.# Instead of: python my_script.py --arg value # Run this: rdg debug python my_script.py --arg value
-
Check your job's output The job output will contain the connection details and the SSH command needed to attach the debugger.
--- Python Debugger Info --- Node: uc2n805.localdomain Port: 51041 Remote Path: /path/to/your/project -------------------------- To connect from a local VS Code instance, run this on your local machine: ssh -N -L 5678:uc2n805.localdomain:51041 <user@login.hostname> Then, attach the debugger to localhost:5678. Script is paused, waiting for debugger to attach...
-
Connect VS Code Follow one of the two methods below depending on your setup. Once attached, you can set breakpoints and debug as if you were running the code locally.
Debugging Workflow
[!NOTE] For the debugger to work, your VS Code editor must have access to the exact source code that is running on the remote compute node.
- If you are developing locally: Make sure you have an identical copy of the project on your local machine (e.g., by using
git clone).- If you are using VS Code Remote-SSH: You are already viewing the project files on the remote machine, so no extra steps are needed.
Method A: Connecting from your Local Machine
Use this method if you are running your IDE locally and want to connect to the remote cluster.
-
Create an SSH Tunnel You can either copy the
sshcommand directly from your job's output (just replace the<user@login.hostname>placeholder), or use therdg tunnelhelper command as shown below.On your local machine, run the
tunnelcommand using theNodeandPortfrom the job output, along with your cluster's SSH login details.# rdg tunnel <NODE> <REMOTE_PORT> <SSH_LOGIN> rdg tunnel uc2n805.localdomain 51041 username@cluster.hostname.com
This will generate an
sshcommand. Copy, paste, and run it in a new terminal to establish the tunnel. Keep this terminal open. -
Attach Debugger (example with VS Code)
- Open the "Run and Debug" panel in VS Code (Ctrl+Shift+D).
- Select "Python Debugger: Remote Attach (via SSH Tunnel)" from the dropdown and click the play button.
- You will be prompted for:
localTunnelPort: The local port for the tunnel (default is5678).remoteWorkspaceFolder: TheRemote Pathfrom the job output.
Method B: Connecting via VS Code Remote-SSH
Use this method if you are already connected to a remote machine (like a login node) using the VS Code Remote - SSH extension.
- Attach VS Code
- Open the "Run and Debug" panel in VS Code (Ctrl+Shift+D).
- Select "Python Debugger: Attach to Compute Node" from the dropdown and click the play button.
- You will be prompted for:
computeNodeHost: TheNodefrom the job output.computeNodePort: ThePortfrom the job output.
Command Reference
| Command | Description |
|---|---|
rdg rebug python <script> [args...] |
Wraps a Python script to start a debugpy listener and waits for a client to attach. |
rdg init |
Creates or updates .vscode/launch.json with the required debugger configurations. |
rdg tunnel <node> <port> <login> [--local-port <port>] |
Constructs the SSH command to establish a tunnel to the compute node. |
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file remote_debug-0.2.5.tar.gz.
File metadata
- Download URL: remote_debug-0.2.5.tar.gz
- Upload date:
- Size: 8.3 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.1.0 CPython/3.13.7
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
78dd951775f95c7f02e550defc145f8313b66af63474d19435c53108ecc9b24e
|
|
| MD5 |
eb5f971d61cc9dd5ce40d0c7a55d8ae1
|
|
| BLAKE2b-256 |
1efde680507efa9d0967354cf26c675c16fafe0e29c0e8afa24f1e320509bcff
|
File details
Details for the file remote_debug-0.2.5-py3-none-any.whl.
File metadata
- Download URL: remote_debug-0.2.5-py3-none-any.whl
- Upload date:
- Size: 7.0 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.1.0 CPython/3.13.7
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
c71b60ac398af07962f5af3480b56a01298bdd0ee700fa13c2d236e54ac42e1e
|
|
| MD5 |
7a36c0f2d429ee91651ae07031dd02ce
|
|
| BLAKE2b-256 |
366c4a82cf4b7f8263106c7772bcbccbbfde379b5c5bbbee593b07d57b4e69b0
|