DEV Community

fj-dig
fj-dig

Posted on

Super – An API-first, lightweight process orchestrator written in Rust (alternative to PM2/Supervisor)

#rust #supervisor #orchestrator

Hey,

Like many of you, I've spent years managing long-running background processes on VPS and small servers (gateway proxies, background workers, edge collectors, local LLM inference scripts, etc.).

For years, the go-to choices have been PM2 or Supervisor:

  • PM2 brings the entire heavy Node.js runtime, which feels completely overkill on small instances or edge boxes just to keep a few binary scripts alive.
  • Supervisor is stuck in the ini/CLI era with Python overhead, making it notoriously painful to control programmatically from modern CI/CD pipelines or automated agents.
  • systemd is rock-solid, but it requires root, is Linux-only, and dynamically registering services via code has high friction.

To scratch my own itch, I built Super — a zero-dependency, API-first lightweight process orchestrator written purely in Rust.

Architecture: Clean Client-Daemon Split

Instead of a bloated all-in-one binary, Super follows a classic daemon/client architecture compiled into two static Rust binaries:

  • superd (Daemon): The silent background service running as a system daemon. It handles process lifecycle, health checks, automatic crash recovery, log streaming, and exposes a clean REST API & WebSocket.
  • super (CLI): A sleek terminal interface for developers. It doesn't fork processes directly; it's simply an official HTTP API client to superd.

Anything you can do via the CLI can be triggered seamlessly via standard HTTP requests from your CI pipelines, Python/Go scripts, or AI agents.

Highlights

  • Zero Runtime Overhead: Pure static Rust binaries. No Node.js, no Python, single-digit MB memory footprint.
  • Transactional Service Updates: Built-in transactional deployment (backup ➔ verify ➔ switch ➔ auto-rollback on failure) so updating a service doesn't feel like gambling.
  • Declarative Stacks: Manage stacks via simple TOML with dependency chains and health probes via super apply.
  • Built-in Audit Ledger: Complete causal event ledger tracking why a process restarted or changed states.
  • Health Probes & Real-time Logs: Native HTTP/TCP health checks, and live log tailing over WebSockets.

Quick Look

Demo

You can spin up a process in seconds:

# Add and auto-start a service
super add --name demo --autostart python3 -m http.server 8080

# Check status and health
super list

# Stream live logs
super logs demo --tail
Enter fullscreen mode Exit fullscreen mode

Installation

You can grab precompiled binaries for Linux (x86_64/arm64), macOS, and FreeBSD directly from the GitHub Releases page.

Or inspect and run the install script (it automatically checks SHA-256 and registers the system service):

curl -fsSL https://github.com/schiplat/super/releases/latest/download/install.sh | sh
Enter fullscreen mode Exit fullscreen mode

(Detailed manual setup guide is available in our Getting Started Docs).

Licensing & Road Ahead

The core (superd engine + super CLI) is 100% open source under the MIT License, completely free, and will stay that way without arbitrary limits. We also have an extensible plugin system in the works for advanced enterprise setups (like multi-tenant RBAC, Web Dashboard, and cgroup isolation).

The project is currently in public beta. I’d love to hear your thoughts, feedback, architectural critiques, or any edge-cases you run into!

Top comments (0)