wait for infrastructure to be build before bringing up workload

[rpuserh]

We have infrastructure that we need to build first which takes 15 min, then we create our app deployment which will use infrastructure

I tried to use fluxv2 Kustomization health assessment In the example below I used Kustomization to check if the workspace finished applying and then deploy our app. But looks like workspace CRD is not compatible with kstatus (unless I did something wrong)

---
apiVersion: kustomize.toolkit.fluxcd.io/v1beta2
kind: Kustomization
metadata:
  name: infra-my-test-env-1
  namespace: terraform-operator
spec:
  interval: 5m
  path: ./terraform-operator/kustomization # <- contains workspace
  prune: true
  sourceRef:
    kind: GitRepository
    name: flux-system
    namespace: flux-system
  healthChecks:
    - apiVersion: app.terraform.io/v1alpha1
      kind: Workspace
      name: my-test-env-1
      namespace: terraform-operator
---
apiVersion: source.toolkit.fluxcd.io/v1beta2
kind: GitRepository
metadata:
  name: podinfo
  namespace: terraform-operator
spec:
  interval: 30s
  ref:
    branch: master
  url: https://github.com/stefanprodan/podinfo
---
apiVersion: kustomize.toolkit.fluxcd.io/v1beta2
kind: Kustomization
metadata:
  name: podinfo
  namespace: terraform-operator
spec:
  interval: 5m0s
  path: ./kustomize
  prune: true
  sourceRef:
    kind: GitRepository
    name: podinfo
    namespace: terraform-operator
  dependsOn:
    - name: infra-my-test-env-1
  targetNamespace: terraform-operator

Helm will be more preferred way for me. So I tried to use helm release. where I create helm chard which will template workspace and defined dependency (so my app helm will wait for infra creation first) but helm applies and it is not waiting for apply to complete. See example below

apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
  name: redis-my-infra-v2
  namespace: terraform-operator
spec:
  interval: 5m
  chart:
    spec:
      chart: terraform-operator/helm/tf-infra
      sourceRef:
        kind: GitRepository
        name: flux-system
        namespace: flux-system
      interval: 1m
  upgrade:
    remediation:
      remediateLastFailure: true
---
apiVersion: source.toolkit.fluxcd.io/v1beta2
kind: HelmRepository
metadata:
  name: podinfo
  namespace: terraform-operator
spec:
  interval: 5m0s
  url: https://stefanprodan.github.io/podinfo
---
apiVersion: helm.toolkit.fluxcd.io/v2beta1
kind: HelmRelease
metadata:
  name: frontend
  namespace: terraform-operator
spec:
  interval: 5m
  chart:
    spec:
      chart: podinfo
      version: ">=4.0.0 <5.0.0"
      sourceRef:
        kind: HelmRepository
        name: podinfo
        namespace: terraform-operator
      interval: 1m
  dependsOn:
    - name: redis-my-infra-v2
  upgrade:
    remediation:
      remediateLastFailure: true
  values:
    backend: http://backend-podinfo:9898/echo
    resources:
      requests:
        cpu: 100m
        memory: 64Mi

The question is if there is a way to tell helm to wait for a workspace to apply.

Or is there another approach that I can take?

[rpuserh]

Any feedback on this ?