v1.33.0

Release v1.33.0

Highlights

• Kubernetes brownfield adoption - terraform import and drift/deletion
  detection are now available for all 17 Kubernetes resources.
• Fix: portainer_stack Kubernetes deployments no longer fail with
  Object not found (key=0) after a successful create (#130).

Added

• Read + Importer implemented for all 17 Kubernetes resources: application,
  clusterrole, clusterrolebinding, configmap, cronjob, helm, ingress,
  job, namespace, namespace_system, role, rolebinding, secret,
  service, serviceaccount, storage, volume. This enables importing
  existing infrastructure and detects out-of-band deletion (404 → resource is
  recreated on the next plan). A shared k8sConfirmExistsByGET helper backs the
  13 manifest-based resources.

Fixed

• portainer_stack with deployment_type = "kubernetes" failed during the
  finalize step with 404 Object not found (key=0) when Portainer's create
  response did not include a usable stack Id (observed on Portainer 2.21.x).
  The provider now resolves the real stack ID by name + endpoint before
  finalizing.

Changed

• Bump go.mongodb.org/mongo-driver 1.17.6 -> 1.17.7.

Upgrade notes

• Read for the Kubernetes resources intentionally does not refresh the
  authored manifest / values / spec fields (the K8s API returns
  server-expanded objects that would cause permanent diffs and, since Update is
  delete+recreate, churn workloads). After terraform import, set those fields
  in config to match the live object, otherwise the next apply will recreate it.
  Per-resource import IDs and the fields you must set are documented in each
  resource's docs page.