Update PodDisruptionBudget api version to get rid of warning message

[wangshu3000]

What this PR does / why we need it:

Update PodDisruptionBudget api version from policy/v1beta1 to policy/v1, to get rid of the warning message, the v1beta1 is deprecated in v1.21+, unavailable in v1.25+;

Which issue this PR fixes

(optional, in fixes #<issue number>(, fixes #<issue_number>, ...) format, will close that issue when PR gets merged)

• fixes Avoid deprecated API versions on YAML definitions for Kubernetes 1.21 #436

Special notes for your reviewer:

Checklist

[Place an '[x]' (no spaces) in all applicable fields. Please remove unrelated fields.]

• Documentation has been updated with helm-docs (run: .github/helm-docs.sh)
• Chart Version bumped
• CHANGELOG.md has been updated
• Variables are documented in the README.md

[clamoriniere]

Hi @wangshu3000

Thanks for this contribution, everything looks good

[rslinckx]

We are using helm to install datadog into our cluster, but we use the helm template feature (part of a gitops setup). As such we are limited in what kind of capabilities the helm command will accurately report, as it does not query the live cluster.

There is an easy way to add capabilities via a command line flag, but there is currently no way to remove capabilities from the default list, which is built-in to helm based on their current k8s library version.

In this case with the latest helm 3 release, the policy/v1 capability seems baked in, and I see no way to disable the capability. As such the helm chart will render a policy/v1 PodDisruptionBudget, which isn't available in my 1.19 cluster.

Would you be open to adding a more explicit flag to select the version of the disruptionbudget to be used in values.yaml ?

[clamoriniere]

Hi @rslinckx
You can set the kubernetes api-server version when running “helm template”

helm template --kube-version 1.19.0 -f values.yaml datadog/datadog

We are using this option to validate our template against several kubernetes versions see:

helm-charts/.github/kubeval.sh

Line 23 in f1820b7

helm dep up "${CHART_DIR}" && helm template --kube-version "${KUBERNETES_VERSION#v}" --values "${CHART_DIR}"/ci/kubeval.yaml "${CHART_DIR}" | ./kubeval --strict --ignore-missing-schemas --kubernetes-version "${KUBERNETES_VERSION#v}" --schema-location "${SCHEMA_LOCATION}"

Please let us know if it solves your issue?

Else we can discuss how we can provide an option in the values.yaml to for a specific version.

[rslinckx]

We are actually running this with helm 3.7.2:

helm template --api-versions apps/v1 --api-versions apiextensions.k8s.io/v1/CustomResourceDefinition --kube-version 1.19 "$chartName" . --namespace "$chartNamespace" --include-crds --no-hooks

but it generates policy/v1 PodDisruptionBudget instead of the expected policy/v1beta1 as 1.19 does not have v1 yet.

I think this is due to helm having some kind of hard-coded k8s versions list based on the version they build against?

I see the CI here uses helm 3.6.3, maybe that's why it worked ? 3.7 updated their internal package to k8s 1.22, where in 3.6 they used k8s 1.21, although it seems v1 was marked as stable in 1.21 already, so not sure this is it...

[clamoriniere]

Hello again @rslinckx
I was able to reproduce your issue too. And I found several open issue on other helm charts reporting the same issue with policy/v1 (argoproj/argo-cd#7291, cortexproject/cortex-helm-chart#277, elastic/helm-charts#1454)

I have created this PR #498 to be more restrictive and fix the issue with helm template

[rslinckx]

Thanks a lot !

[mscanlon72]

I am using EKS v1.21 and I still see the warning messages using the latest charts (datadog-2.28.13). Any information on why that would be the case?

[wangshu3000]

@mscanlon72
Did you see the warning when running
kubectl get PodDisruptionBudget
or saw the warning when deploying the chart?
I think you may still see a warning when you run kubectl get PodDisruptionBudget.
Kubernetes 1.21 still support policy/v1beta1, it'll be removed after 1.25.

[mscanlon72]

This is what I see when I run that command:

No resources found in datadog namespace.

[mscanlon72]

Setting createPodDisruptionBudget to true creates the PDB, but the warnings still persist in the logs and are very frequent.