Provider crash with 'Plugin did not respond' error on complex kubernetes_manifest resources

[maateen]

Bug Report

Environment

• Provider Version: v2.31.0
• Terraform/OpenTofu Version: OpenTofu v1.10.3
• Platform: Darwin arm64 (macOS Sequoia 24.5.0)
• Go Version: go1.24.2 darwin/arm64

Issue Description

The kubernetes provider crashes with "Plugin did not respond" when planning kubernetes_manifest resources containing complex nested objects (ArgoCD ApplicationSets).

Error Message

Error: Plugin did not respond

  with kubernetes_manifest.applicationset,
  on applicationset.tf line 2, in resource "kubernetes_manifest" "applicationset":
   2: resource "kubernetes_manifest" "applicationset" {

The plugin encountered an error, and failed to respond to the
plugin.(*GRPCProvider).ReadResource call. The plugin logs may contain more
details.

Stack Trace

panic: ElementKeyInt(0): can't use tftypes.Object["clusterDecisionResource":tftypes.Object["configMapRef":tftypes.String, "labelSelector":tftypes.Object["matchExpressions":tftypes.List[tftypes.Object["key":tftypes.String, "operator":tftypes.String, "values":tftypes.List[tftypes.String]]], "matchLabels":tftypes.Map[tftypes.String]], "name":tftypes.String, "requeueAfterSeconds":tftypes.Number, "template":tftypes.Object[...]]

Configuration

The failing resource is a kubernetes_manifest for an ArgoCD ApplicationSet with complex nested generators containing matrix configurations, git directories, and cluster definitions.

Provider Configuration

Using EKS authentication via AWS CLI exec plugin:

provider "kubernetes" {
  host = "${dependency.eks.outputs.cluster_endpoint}"
  cluster_ca_certificate = base64decode("${dependency.eks.outputs.cluster_certificate_authority_data}")
  
  exec {
    api_version = "client.authentication.k8s.io/v1beta1"
    command = "aws"
    args = ["eks", "get-token", "--role-arn", "${local.vars.role_arn}", "--cluster-name", "${dependency.eks.outputs.cluster_name}"]
  }
}

Analysis

This appears to be a tftypes handling issue in the provider when processing complex nested object structures in kubernetes_manifest resources. The panic occurs during the ReadResource call, suggesting the provider cannot properly handle the complex nested ApplicationSet specification.

Reproduction

The issue occurs consistently when running terraform plan or terragrunt plan on configurations containing ArgoCD ApplicationSet manifests with complex nested generator configurations.

[fatmcgav]

#2800 should fix this one 🤞