Back to catalog
/dev/push
MIT/dev/push is a self-hosted PaaS alternative to Vercel that deploys Python and Node.js applications directly from GitHub onto your own servers.
Open-source alternative to — compare it against:
Overview
Small engineering teams running Python backends often face an awkward trade-off: pay recurring per-seat fees on managed platforms or spend days configuring continuous deployment pipelines. /dev/push resolves that operational tension by turning an ordinary Linux server into a self-hosted deployment hub. Code pushed to GitHub triggers zero-downtime container updates, isolated preview URLs, and automated SSL certificates without routing traffic through third-party infrastructure.
Why /dev/push instead of Vercel
Vercel built its reputation on frictionless web delivery: connect a repository, push a branch, and receive a working preview URL backed by edge routing. For Next.js frontends and brief serverless invocations, that managed workflow is fast. However, teams running persistent backend services—FastAPI workers, Django APIs with database connection pools, or long background jobs—hit serverless execution timeouts and steep bandwidth pricing. Vercel is designed around ephemeral edge handlers, not long-running stateful containers.
/dev/push replicates that push-to-deploy experience directly on an Ubuntu or Debian VPS. Standard Docker containers run behind Traefik, giving operations teams direct SSH access, predictable server costs, and granular memory limits per project. You trade away Vercel's global edge network and managed data services, but you gain unmetered execution time, data privacy, and support for any language runtime that packages into a container.
What it does well
- Git-driven rollouts: Pushing to GitHub triggers automated builds, zero-downtime updates, and isolated preview environments per branch.
- Instant rollback: Operators can redirect an environment alias to any previous successful build without waiting for a rebuild.
- Automatic HTTPS: Traefik integration and Let's Encrypt handle wildcard routing and SSL certificates for branches and custom apex domains.
- Multi-runtime Docker support: Native presets package Python (FastAPI, Flask, Django) and Node.js applications, while custom containers support other stacks.
- Encrypted secrets: Project environment variables are stored with Fernet encryption and injected directly into runtime containers.
- Live streaming logs: Real-time build and runtime logs stream over server-sent events to accelerate operational debugging.