Hi, I'm Burak, the developer of RunR.

RunR is software you install, not a hosted service. The control plane comes as a Helm chart (one required value: your domain) or as a Docker Compose stack for a single host, air-gapped hosts included. Then you register any Kubernetes cluster as a region and RunR installs Knative Serving, Traefik, a registry, Redis, Prometheus and Loki there.

From then on you write a function in the browser, git push a project, or bring a container image or a Wasm module, and RunR builds it and runs it with scale-to-zero. Every code or configuration change is a version you can deploy, compare and roll back with the configuration it ran with. Projects have dev, staging and production environments, blue-green and canary rollouts with automatic rollback, and can run in several regions behind one host name.

A few things that are less common in this space: each project type of a region runs under runc, gVisor or Kata Containers; every language is a plugin (Python, Node.js, Go, PHP, PyTorch and TensorFlow come built in); GPU workers and Hugging Face model services; a built-in MCP server so agents can operate RunR within their token's permissions; and a Sandbox API for code that agents write. RBAC, SSO, an audit log, secrets that stay in the region and signed-image admission are included; multi-tenancy is an optional module.

Try it without installing anything: https://demo.runr.sh (credentials are on the sign-in page). Docs: https://runr.sh/docs/ - code: https://github.com/BRF-Tech/runr

I'd like to hear where getting a first region up took longer than it should have.

Hi, I'm Burak, the developer of RunR.

Try it without installing anything: https://demo.runr.sh (credentials are on the sign-in page). Docs: https://runr.sh/docs/ - code: https://github.com/BRF-Tech/runr

I'd like to hear where getting a first region up took longer than it should have.

RunR

Self-hosted serverless on Kubernetes and Knative

Self-hosted serverless on Kubernetes and Knative