refunc

A team with functions written around a Lambda-style invocation interface may want to run those functions on its own Kubernetes infrastructure. Refunc provides a function platform with runtime adapters and a Lambda-compatible API for that purpose. The familiar interface does not reproduce every AWS service integration or execution guarantee. Its documented examples are historical, so verify current runtime and Kubernetes support before relying on it, then test the actual event, retry and timeout behavior required by the application.

Current guidance

The recovered resource is refunc/refunc, a Kubernetes function platform with a Lambda-compatible API, runtime adapters and NATS-based transport. The available README still demonstrates a Python 3.7 runtime and a minimal play environment. Those examples identify the project but do not establish a supported current runtime or Kubernetes deployment.

Compatibility with the shape of a Lambda API is narrower than compatibility with all AWS execution semantics. Before adoption, verify event sources, IAM expectations, runtime lifecycle, concurrency, retries, timeouts and observability for the actual function. A workload that depends on managed AWS integrations may require application changes even if an invocation succeeds through a familiar client.

For an existing installation, preserve function bundles, runtime images, trigger definitions and the state of any supporting message or object services. Ask the maintainer for a current supported release path, then run contract tests with representative failures and duplicate events in isolation. If that evidence cannot be obtained, evaluate a maintained execution platform and migrate explicit interfaces rather than treating old scale-to-zero claims as validated current behavior.