chart2kro
Documentation menu

Getting started

Install chart2kro, convert the podinfo Helm chart into a kro API, and run two instances of it on a local cluster in about ten minutes.

You will convert the podinfo chart into a new Kubernetes API, Podinfo, and create two instances of it. New to kro? Skim Background first.

1. Install chart2kro

brew install hupe1980/tap/chart2kro
# or
go install github.com/hupe1980/chart2kro/cmd/chart2kro@latest

Or download a signed release (how to verify it), or run the image: docker run --rm ghcr.io/hupe1980/chart2kro --help. Converting needs no cluster, Helm or kubectl.

2. Get a cluster with kro

Skip this if you already have kro v0.9.2 or newer.

kind create cluster
helm install kro oci://registry.k8s.io/kro/charts/kro --version 0.9.4 \
  --namespace kro-system --create-namespace --wait

3. Convert the chart

chart2kro convert oci://ghcr.io/stefanprodan/charts/podinfo --version 6.15.0 -o podinfo/
podinfo 6.15.0 → Podinfo.kro.run/v1alpha1: 3 resources, 29 exposed, 33 frozen, 0 failed, 60 unused (256 witnesses, kro v0.9.4, Kubernetes 1.37)
  frozen: changes-structure             15
  frozen: toggles-fields                8
  frozen: transformed                   6
  frozen: unknown-kind                  4
  unused: affects-only-dropped-objects  11
  run 'chart2kro inspect' for details
  dropped: Pod <generated> (test hook)

The summary says: the chart has 3 resources; 29 values became inputs of the new API; the rest stay at their defaults, and each has a reason. The 256 witnesses are the value combinations at which kro's output was compared with Helm's.

podinfo/ now holds:

FileContents
rgd.yamlThe ResourceGraphDefinition: the API's schema and the resource templates
instance.yamlAn example instance with every input at its default
report.jsonEvery value and resource, with its outcome and reason
prerequisites.yamlOnly if the chart has CRDs or shared cluster objects: apply it once, first

report.json follows a published JSON schema.

4. Install the API

kubectl apply -f podinfo/rgd.yaml
kubectl get rgd podinfo
NAME      APIVERSION   KIND      STATE    READY   AGE
podinfo   v1alpha1     Podinfo   Active   True    2s

kro has created the Podinfo kind. Its schema is the chart's values, typed and defaulted.

5. Create instances

The example instance uses every default:

kubectl apply -f podinfo/instance.yaml

A second instance sets two inputs. Fields mirror values.yaml:

apiVersion: kro.run/v1alpha1
kind: Podinfo
metadata:
  name: blue
spec:
  replicaCount: 2
  ui:
    color: "#34577c"
kubectl get podinfo
kubectl get deploy
NAME      STATE    READY   AGE
blue      ACTIVE   True    36s
podinfo   ACTIVE   True    36s

NAME              READY   UP-TO-DATE   AVAILABLE   AGE
blue-podinfo      2/2     2            2           36s
podinfo-podinfo   1/1     1            1           36s

Each instance gets its own objects, named <instance>-podinfo, and its own pods: chart2kro added an app.kubernetes.io/instance selector label so the two never share a Service.

Delete an instance and kro deletes its objects: kubectl delete podinfo blue.

6. Find out why a value is not an input

chart2kro inspect oci://ghcr.io/stefanprodan/charts/podinfo --version 6.15.0 --value ui.message
ui.message: frozen
  reason: changes-structure
  where:  Deployment (podinfo/templates/deployment.yaml) spec.template.spec.containers[0].env
  the value stays at its default in every instance

Setting ui.message adds an environment variable, so it changes the structure of the Deployment, which a kro expression cannot do. Set it as a platform default with -f instead.

Next steps