PlatformEngr® Progressive delivery course

See the end result · about 30 minutes

A canary that rolls itself back: a preview of where the course takes you

On a scratch kind cluster, install Argo Rollouts and a small Prometheus, release checkout as a canary judged by its own success rate, and watch a good release walk through its steps alone. Then ship a release that fails one request in three and watch it abort with nobody watching.

This is a preview, not the first step. It installs everything with Helm by hand and explains little, so you can see where you are heading. The full course does it on your Argo CD platform from the Argo CD course, through Git, and explains every decision.

This preview

A scratch cluster, Helm by hand, one canary. Commands to copy, few explanations.

The full course

10 modules on your Argo CD platform: Rollouts from Git, blue-green, analysis, a self-rolling-back canary, Istio, Gateway API, team release actions and a failed-release drill.

Your setup

Every command below uses these values. They stay in this browser and are shared with the full course. The preview uses its own kind cluster, rollouts-preview, so your Argo CD cluster is not touched.

    What just happened

    new image tag ── Rollout ── 1 of 4 pods on the new version │ Prometheus ── success rate of the new pods only ── AnalysisRun │ 95% or more: next step ── 50% ── 75% ── promoted less, twice: abort ── every pod back on the stable version

    The full course builds this on your Argo CD platform, through Git, and adds blue-green, exact weights with Istio, Gateway API and the buttons the product team gets.