Linkerd rebuilds its continuous integration (CI)

With Kubernetes In Docker (KinD) and GitHub Actions

In this 2020 KubeCon EU presentation, our own Andrew Seigner shares how we migrated our continuous integration (CI) system built on a GKE cluster and Travis to a system running eight one-off Kubernetes in Docker (KinD) clusters in parallel via GitHub Actions.

Running all serialized tests on a single Kubernetes cluster took the legacy CI system about 45 minutes followed by an often multi-hour backup. Linkerd’s CI system today is parallelizable and runs in less than ten minutes – a huge improvement!

This talk covers Linkerd’s CI journey from a single, persistent Kubernetes cluster to theoretically unlimited one-off KinD clusters. You’ll learn what patterns and tools worked well (and not-so-well) for us.

You’ll also see two demos of Linkerd’s CI workflow, including automated cluster setup, test, and teardown.