Custom Resource Definitions

This section of the Best Practices Guide deals with creating and using Custom Resource Definition objects.

When working with Custom Resource Definitions (CRDs), it is important to distinguish two different pieces:

  • There is a declaration of a CRD. This is the YAML file that has the kind CustomResourceDefinition
  • Then there are resources that use the CRD. Say a CRD defines foo.example.com/v1. Any resource that has apiVersion: example.com/v1 and kind Foo is a resource that uses the CRD.

Install a CRD Declaration Before Using the Resource

Helm is optimized to load as many resources into Kubernetes as fast as possible. By design, Kubernetes can take an entire set of manifests and bring them all online (this is called the reconciliation loop).

But there's a difference with CRDs.

For a CRD, the declaration must be registered before any resources of that CRDs kind(s) can be used. And the registration process sometimes takes a few seconds.

Method 1: Let helm Do It For You

With the arrival of Helm 3, we removed the old crd-install hooks for a more simple methodology. There is now a special directory called crds that you can create in your chart to hold your CRDs. These CRDs are not templated, but will be installed by default when running a helm install for the chart. If the CRD already exists, it will be skipped with a warning. If you wish to skip the CRD installation step, you can pass the --skip-crds flag.

Some caveats (and explanations)

There is no support at this time for upgrading or deleting CRDs using Helm. This was an explicit decision after much community discussion due to the danger for unintentional data loss. Furthermore, there is currently no community consensus around how to handle CRDs and their lifecycle. As this evolves, Helm will add support for those use cases.

The --dry-run flag of helm install and helm upgrade is not currently supported for CRDs. The purpose of "Dry Run" is to validate that the output of the chart will actually work if sent to the server. But CRDs are a modification of the server's behavior. Helm cannot install the CRD on a dry run, so the discovery client will not know about that Custom Resource (CR), and validation will fail. You can alternatively move the CRDs to their own chart or use helm template instead.

Another important point to consider in the discussion around CRD support is how the rendering of templates is handled. One of the distinct disadvantages of the crd-install method used in Helm 2 was the inability to properly validate charts due to changing API availability (a CRD is actually adding another available API to your Kubernetes cluster). If a chart installed a CRD, helm no longer had a valid set of API versions to work against. This is also the reason behind removing templating support from CRDs. With the new crds method of CRD installation, we now ensure that helm has completely valid information about the current state of the cluster.

Method 2: Separate Charts

Another way to do this is to put the CRD definition in one chart, and then put any resources that use that CRD in another chart.

In this method, each chart must be installed separately. However, this workflow may be more useful for cluster operators who have admin access to a cluster