Running local dev processes: from terminal tabs to a process manager

Judge any setup by four questions. Where does each server’s output go, and can you read it after the terminal closes? What happens when one process crashes? What happens when two want the same port? Can a teammate, or you next Monday, start the whole set without remembering the order? Run multiple dev servers locally answers them for terminal tabs, tmux, concurrently, Procfile runners, Docker Compose, and Rig in one comparison table.

• A process manager starts programs, tracks them, captures their output, and decides what happens when they exit. A task runner finishes when its command does, and a viewer such as top does not own the processes it lists.
• A Procfile lists one process per line as a name, a colon, and a command. Heroku defines the format and other runners read it, so each tool decides for itself what it accepts. The Procfile guide covers syntax and runners, and the Procfile validator checks a file line by line.
• A sidecar process is a helper that runs beside the main app, such as a queue worker. Nothing ties it to the web server except whoever starts both.
• A health check is a small, repeatable request that says whether a service is alive and able to work. Listening on a port is not the same as being healthy.
• A reverse proxy accepts requests on behalf of a backend. Locally, it often puts HTTPS in front of a plain HTTP dev server.

1. Read the comparison guide to choose an approach.
2. If you choose a Procfile, read the Procfile guide, then check your file with the Procfile validator.
3. Add health checks once the start order matters, and read the sidecar process entry if you run workers.
4. Ports are the usual next problem, so continue with ports and conflicts.

Rig is a process manager for local dev servers. It detects package.json scripts, a Procfile, and a Compose file in a project and offers each as a script to start. It keeps one log per project, polls health over HTTP, and, when a project has auto-restart switched on, restarts a crashed process with backoff. Procfile projects, Health-aware start order, and the Mac process manager page describe it.

A Procfile only says what to run: it installs nothing and builds nothing, and runners differ on what happens when one process exits, so kill a worker once and watch. Compose runs your code in a container, so live editing needs a bind mount or Compose Watch. Rig’s health check is an HTTP request, so a worker with no URL reads as unknown, and a 404 counts as healthy.