User Guide: Using OpenCode on HPC

OpenCode runs in your terminal and talks to a shared Ollama server on node01 — no cloud API, no API keys. For installation, provider configuration, TUI commands, and general usage, see OpenCode on Your Laptop — the setup is identical whether you are on node02 or your own machine.

This page covers the one HPC-specific requirement: all OpenCode sessions must run inside a SLURM allocation.

warning

Processes left running outside a SLURM allocation for more than 5 minutes will be automatically terminated.

---

1. Request a SLURM Allocation

Before launching OpenCode, request an interactive allocation:

srun --time=04:00:00 --cpus-per-task=2 --mem=4G --pty bash

• No GPU request needed — inference runs on node01's GPUs over the network
• --cpus-per-task=2 and --mem=4G are sufficient for OpenCode alone
• Adjust --time as needed; 4 hours is a reasonable working window

If you plan to compile code or run tests in the same session, request more:

srun --time=04:00:00 --cpus-per-task=8 --mem=16G --pty bash

Your shell prompt will show slurm-<jobid> while inside the allocation.

2. Launch OpenCode

cd ~/your_project
opencode

3. End the Session

Exit OpenCode first, then release the allocation:

/exit

exit

---

Troubleshooting

• OpenCode got killed or you see a termination warning: you ran OpenCode outside a SLURM allocation — restart from Step 1.
• For all other issues (provider config, /models 404, connectivity), see the Troubleshooting section in the laptop guide.