fix: rolling back PHS creation in deployment (#105)

Author: bradmiro

dataproc.tf

@@ -103,30 +103,3 @@ resource "google_project_iam_member" "bq_connection_iam_biglake" {
   role    = "roles/biglake.admin"
   member  = "serviceAccount:${google_bigquery_connection.ds_connection.cloud_resource[0].service_account_id}"
 }
-
-resource "google_dataproc_cluster" "phs" {
-  name    = "gcp-${var.use_case_short}-phs-${random_id.id.hex}"
-  project = module.project-services.project_id
-  region  = var.region
-  cluster_config {
-    staging_bucket = google_storage_bucket.phs-staging-bucket.name
-    temp_bucket    = google_storage_bucket.phs-temp-bucket.name
-    gce_cluster_config {
-      service_account = google_service_account.dataproc_service_account.email
-      subnetwork      = google_compute_subnetwork.subnet.name
-    }
-    software_config {
-      override_properties = {
-        "dataproc:dataproc.allow.zero.workers" = "true"
-        "spark:spark.history.fs.logDirectory"  = "gs://${google_storage_bucket.spark-log-directory.name}/phs/*/spark-job-history"
-      }
-    }
-    endpoint_config {
-      enable_http_port_access = "true"
-    }
-  }
-
-  depends_on = [
-    google_project_iam_member.dataproc_sa_roles
-  ]
-}

test/integration/analytics_lakehouse/analytics_lakehouse_test.go

@@ -47,8 +47,8 @@ func TestAnalyticsLakehouse(t *testing.T) {
 
 		verifyNoVMs := func() (bool, error) {
 			currentComputeInstances := gcloud.Runf(t, "compute instances list --project %s", projectID).Array()
-			// There should only be 1 compute instance (Dataproc PHS). Wait to destroy if other instances exist.
-			if len(currentComputeInstances) > 1 {
+			// If compute instances is greater than 0, wait and check again until 0 to complete destroy
+			if len(currentComputeInstances) > 0 {
 				return true, nil
 			}
 			return false, nil

workflows.tf

@@ -170,13 +170,3 @@ resource "time_sleep" "wait_after_all_workflows" {
     data.http.call_workflows_project_setup,
   ]
 }
-
-# Stop the PHS cluster after creation since it costs too much.
-# tflint-ignore: terraform_unused_declarations
-data "http" "call_stop_cluster" {
-  url    = "https://dataproc.googleapis.com/v1/projects/${module.project-services.project_id}/regions/${var.region}/clusters/${google_dataproc_cluster.phs.name}:stop"
-  method = "POST"
-  request_headers = {
-    Accept = "application/json"
-  Authorization = "Bearer ${data.google_client_config.current.access_token}" }
-}