tkn pipelinerun logs $PIPELINE_RUN_NAME --follow returns: "unable to decode an event from the watch stream: unable to decode watch event: no kind \"PipelineRun\" is registered for version \"tekton.dev/v1\" about tekton-argocd-eks

[jonashackt]

Seems to be related to tektoncd/pipeline#5964

This PipelineRun was originally created as v1beta1, but it is being converted to v1 and I'm not sure why.

That's exactly the case, looking into the PipelineRun it now has another version tekton.dev/v1 instead of the original file, which has tekton.dev/v1beta1.

$ tkn pipelinerun describe $PIPELINE_RUN_NAME -o yaml

apiVersion: tekton.dev/v1
kind: PipelineRun
metadata:
  annotations:
    kubectl.kubernetes.io/last-applied-configuration: |
...

Ah seems to be the case only when starting the PipelineRun with kubectl create:

...when the pipelinerun is created with kubectl create, kubectl is specifying somehow that a v1 PipelineRun should be used. There's a bit more detail in kubernetes/kubernetes#58131: it says that "kubectl get deployment foo is ambiguous, since the server has deployments in multiple api groups. When a resource exists in multiple api groups, kubectl uses the first group listed in discovery docs published by the server which contains the resource."

[jonashackt]

Oh my... it seems to be an issue of the current release architecture of Tekton itself - they are slowly moving from tekton.dev/v1beta1 to tekton.dev/v1 and don't have everything coordinated well: see tektoncd/plumbing#413 (comment) stating:

The lack of well-coordinated versions resulted in multiple issues in Pipelines v0.43.x

Digging around a bit I also found tektoncd/pipeline#5932 (comment) where one solution is

Downgrading your pipelines version should also fix this problem.

Well that's really kind of annoying, but I locally tested it and using version v0.42.0 instead of v0.45.0 solves the problem.