Use one operator to manage multiple CRDs

[gajananbhat]

How do I use one operator to manage multiple crds?

@kopf.on.create('r1.com', 'v1', 'power')
@kopf.on.create('r1.com', 'v1', 'temp')
def create_fn(meta, spec, namespace, logger, body, kind, **kwargs):

The above doesn't work.

Also, how do I get the kind of Custom Resource inside the create_fn? "kind" argument does not work.

Thanks

[nolar]

It should work, you are doing it right. Can you please explain how exactly that does not work? Specifically, the setup and code snippets (CRD yamls) so that I could reproduce the case where it does not work.

To know the resource, there is a kwarg named resource. This is a dataclass with fields representing all known information about the resource (as taken from the initial cluster scanning). So, resource.kind can be used to get the kind.

• https://kopf.readthedocs.io/en/stable/kwargs/#resource-related-kwargs
• https://kopf.readthedocs.io/en/stable/packages/kopf/#kopf.Resource

Also, you can use param to distinguish several decorators/causes in the same function.

• https://kopf.readthedocs.io/en/stable/kwargs/#parametrization

[gajananbhat]

Hi @nolar,

Apologies for the delay. Yes. It works now. It was a typo on my part that caused the confusion.

Also, thanks for the pointer to "resource.kind". I am now able to process CRs based on their kind.

How can we return messages to kubectl output in case of any failure during creation/deletion?

Thanks

[nolar]

There is no way to return messages to kubectl with create/update/delete handlers. They are invited after everything has happened.

But you can take a look at admission hooks, which can either reject the creation/update (with a message), or accept/modify the object and yield some warnings to the user.

[gajananbhat]

Thank you. Is there a way to integrate custom resource logs with "kubectl logs" output? I can see the output if I do "kubectl describe" on a custom resource. But, it would be great if we can get it with kubectl logs.