5 months ago I built a VS Code extension to manage SLURM jobs. Since then, it’s evolved into a full cluster management tool.
Hey everyone,
About 5 months ago, I posted here about a side project I was working on: sCode, a VS Code extension to manage SLURM jobs directly from the editor.
Initially, it was just a simple way to avoid typing squeue over and over. But based on a lot of my own workflow needs and some great feedback, it has evolved into a much more comprehensive cluster management tool over the last few months.
I’ve essentially tried to turn VS Code into a unified control center for HPC work so you never have to context-switch to a terminal while working on your scripts.
Here are the major updates since the first version:
- Live GPU Monitoring: Added a dedicated view that uses
nvidia-smito show GPU partition usage, memory stats, and queue pressure. - The "Hall of Shame": A fun leaderboard feature that ranks the cluster’s top GPU hogs (with emojis like 🐷 Job Hog and 🧛 VRAMpire).
- One-Click Job Arrays: You can now cancel specific indices or ranges within a job array without nuking the whole array.
- Smart Log Resolution: Right-click any active or historical job in the sidebar to instantly open its
stdoutorstderrfile. - Quick Submit with Dependencies: A ▶ button in your
.shscripts to submit immediately, plus a guided UI for setting upafterokorafteranydependencies. - And many more features....
If you work on a cluster and use VS Code Remote, I'd love for you to give the new version a try and let me know what you think. What features would you need to make this a daily driver for your workflow?
GitHub Repo:https://github.com/dhimitriosduka1/sCode
OpenVSX: https://open-vsx.org/extension/DhimitriosDuka/slurm-cluster-manager
Marketplace: https://marketplace.visualstudio.com/items?itemName=DhimitriosDuka.slurm-cluster-manager