Added supervisord to survive process restart

[mishushakov]

Note

Medium Risk
Changes the container/service startup and supervision model, which can affect sandbox availability and restart behavior if misconfigured. Added end-to-end tests mitigate risk but failures could still impact runtime stability.

Overview
Replaces the previous start-up.sh flow with supervisord, introducing supervisord.conf plus dedicated start-jupyter.sh/start-code-interpreter.sh wrappers (including a PID file and a “wait for Jupyter ready” gate) so both services are automatically restarted together.

Updates the template build to install supervisor and ship the new scripts/config, and adds JS + Python (sync/async) integration tests that forcibly kill -9 Jupyter or the code-interpreter process and assert /health and code execution recover.

^{Written by Cursor Bugbot for commit 35f17c9. This will update automatically on new commits. Configure here.}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c36ff2959f

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

[cursor]

Unguarded PID file read may produce spurious errors

Low Severity

The kill "$(cat /var/run/code-interpreter.pid)" 2>/dev/null command doesn't check if the PID file exists before reading it. The 2>/dev/null only suppresses kill's stderr, not cat's. If Jupyter exits before code-interpreter has written its PID file (e.g., very early crash on first startup), cat emits an unhandled error to stderr (visible in supervisord logs), and kill receives an empty-string argument. Adding a file-existence guard (e.g., wrapping in [ -f ... ] &&) would make this robust.

 

[ValentaTomas]

As we already have systemd inside of the sandbox, don't we want to use that for this too?

[mishushakov]

@ValentaTomas but then it's impossible to test code interpreter locally