Documentation for self-hosted deployment on single-node setups

[renezander030]

I'm evaluating e2b infrastructure for running AI agent sandboxes on a single VPS (Debian, 8GB RAM). The repo has great architecture docs for the full cloud setup, but I'd love to see a minimal self-hosted deployment guide.

Specifically:

• Minimum resource requirements for running the orchestrator + a few concurrent sandboxes
• Whether Firecracker microVMs are required or if there's a lighter-weight isolation option for small deployments
• Any gotchas around running this outside of AWS/GCP

Happy to contribute docs if someone can point me at the right entry points.

[linear]

ENG-3576

[jakubno]

Please check local dev guide, it's probably closest to what you want

Whether Firecracker microVMs are required or if there's a lighter-weight isolation option for small deployments

We only support running it with Firecracker

Any gotchas around running this outside of AWS/GCP

It meant for local dev only not, it isn't considered production ready. Some features may be missing.

[renezander030]

Thanks, went through the local dev guide and the compose setup.

One thing I want to confirm: the compose file handles all the infra (db, cache, observability), but the actual application services run separately through Make, right?

I'm going to try getting this running on a single machine this week. If it works I'll open a PR with a minimal getting-started guide. Want me to add to the local dev guide or a separate doc?