Error: Cycle on destroy

[scleriot]

When I make some changes in my kubectl_manifest resource, I get the following error on terraform apply

Error: Cycle: kubectl_manifest.workers (destroy), google_container_cluster.sidekiq_workers, provider["registry.terraform.io/gavinbunney/kubectl"]

Here is my Terraform config:

provider "kubectl" {
  host                   = google_container_cluster.sidekiq_workers.endpoint
  cluster_ca_certificate = base64decode(google_container_cluster.sidekiq_workers.master_auth[0].cluster_ca_certificate)
  token = data.google_client_config.provider.access_token
  load_config_file       = false
}

resource "kubectl_manifest" "workers_1" {
  wait      = true
  yaml_body = yamlencode({
    apiVersion: "apps/v1",
    kind: "Deployment"
    metadata: {
      name: "workers_1"
      namespace: "default"
      labels: {
        app: "workers_1"
      }
    },
    spec: {
      replicas: 1,
      selector: {
        matchLabels: {
          app: "workers_1"
        }
      },
      template: {
        metadata: {
          labels: {
            app: "workers_1"
          }
        },
        spec: {
          containers: [
            {
              name: "worker-1",
              image: local.api_image,
              env: concat(local.cloud_run_envs, [
                { name  = "DB_SOCKET", value = "/var/run/mysqld/mysqld.sock" },
                { name  = "DB_HOST", value = google_sql_database_instance.database.private_ip_address },
                { name = "RUNTIME", value = "worker" }
              ])
            }
          ]
        }
      }
    }
  })
}

If I manually remove the container from GKE web UI, then the deployment works

[w32-blaster]

seeing similar behavior when trying to recreate resources. @scleriot did you find a solution?