Introducing Oxmgr — A Rust Process Manager Built for Developers Who Care About Overhead

We got tired of PM2 eating 80MB of RAM before our app even started.

On a $6/month VPS with 1 GB of memory, that’s 8% of the machine doing nothing but keeping processes alive. Multiplied across dozens of servers, it becomes a meaningful cost. And when something crashed, the 400ms recovery window was visible in our error rates.

So we built Oxmgr — a process manager written in Rust that starts in 38ms and uses 4MB at rest.

What Made Us Build This

The core problem isn’t that PM2 is bad. It’s excellent for what it is: a JavaScript daemon that manages JavaScript processes. But “JavaScript daemon” means it carries the weight of Node.js everywhere it goes.

If you’re not familiar with what a process manager does in the first place, the process manager primer is a good starting point before continuing here.

We wanted something with no runtime, no garbage collector, and no V8 warmup between your processes crashing and restarting. Rust was the obvious choice.

What Oxmgr Does

Oxmgr manages long-running processes through a single config file, oxfile.toml:

[processes.api]
command = "node dist/server.js"
instances = 4
env = { NODE_ENV = "production", PORT = "3000" }
restart_on_exit = true
restart_delay_ms = 100
watch = ["dist"]

[processes.api.health_check]
endpoint = "http://localhost:3000/health"
interval_secs = 30

[processes.worker]
command = "python worker.py"
restart_on_exit = true
max_restarts = 5

oxmgr start brings everything up. oxmgr status shows you what’s running. oxmgr reload does a rolling restart with zero downtime.

It’s not just for Node.js — Oxmgr manages any process: Python scripts, Go binaries, Ruby apps, compiled Rust services.

The Numbers

On identical hardware (AWS t3.small), managing 10 Node.js processes:

The crash recovery number is what we’re most proud of. 11ms is below the threshold of a single HTTP request timeout. Your users don’t see the crash.

What’s Different From PM2

Config file instead of CLI flags. Everything is in oxfile.toml, committed to git, reviewed in PRs. No more wondering what flags the process was started with.

Written in Rust. No Node.js dependency, no npm version conflicts, no pm2 update ceremony. Oxmgr is a single binary.

PM2 migration in one command. If you have an ecosystem.config.js, run:

oxmgr import ecosystem.config.js

It generates an equivalent oxfile.toml for review.

Getting Started

Install via npm:

npm install -g oxmgr

Or with brew on macOS:

brew install oxmgr

Create an oxfile.toml:

[processes.app]
command = "node server.js"
restart_on_exit = true

Start:

oxmgr start

That’s two minutes from zero to crash-recovering production-ready process management.

What’s Coming

We’re actively building:

• TUI dashboard — real-time terminal UI with resource graphs
• Git webhooks — auto-deploy on push without a CI pipeline
• System service integration — register processes as systemd/launchd units with one command
• Resource limits — cap CPU and memory per process, restart on limits exceeded

Follow the project on GitHub. If you try it, open an issue — we want to hear what you’re running and what you need.