📚 / Dockerized Supervision / HowTo: Write a Unit

Create and deploy your own Units


A Unit represents the base element of your supervision, it describes any task that must execute periodically and should succeed every time.

This means that a unit produces at least 2 kinds of data:

  • logs: the textual output of the task being run (ie: curl)
  • state: the final result of the task: was it successful? or did it fail?

Optionally, a unit can produce artifacts (files) thanks to Kubernetes' persistent volumes.

How does it work?

Kubirds relies on TektonCD to schedule your supervision workload:

  • for each Unit resource, a Pipeline resource is created
  • according to the Unit’s schedule, a PipelineRun resource is created periodically

NB: For each Reactor resources, a task is appended to the Pipeline resources.

For more information, see this document.

The Unit resource specify which Docker image to pull to run your task. This means you can provide a black box to Kubirds, as long as it respects the container contract, it will be executed in your Kubernetes cluster.

This workflow has the following advantages:

  • your tasks will run in your cluster seamlessly, no need to deploy anything else
  • your supervision infrastructure scales with your cluster, with no action required
  • any scenario you could imagine can be integrated within Kubirds, only Docker images are required

How to write a Unit?

In the next part of this tutorial, we’ll see: