Let TPU CI read from project GCS buckets (#7600)

pytorch · Jul 1, 2024 · 6873d33 · 6873d33
1 parent 3bdca4d
commit 6873d33
Show file tree

Hide file tree

Showing 2 changed files with 3 additions and 1 deletion.
diff --git a/infra/terraform_modules/arc_v4_container_cluster/main.tf b/infra/terraform_modules/arc_v4_container_cluster/main.tf
@@ -36,6 +36,7 @@ resource "google_container_node_pool" "arc_v4_cpu_nodes" {
     oauth_scopes = [
       "https://www.googleapis.com/auth/logging.write",
       "https://www.googleapis.com/auth/monitoring",
+      "https://www.googleapis.com/auth/devstorage.read_only",
     ]
   }
 
@@ -60,6 +61,7 @@ resource "google_container_node_pool" "arc_v4_tpu_nodes" {
     oauth_scopes = [
       "https://www.googleapis.com/auth/logging.write",
       "https://www.googleapis.com/auth/monitoring",
+      "https://www.googleapis.com/auth/devstorage.read_only",
     ]
     machine_type = "ct4p-hightpu-4t"
   }

diff --git a/infra/tpu-pytorch/tpu_ci.tf b/infra/tpu-pytorch/tpu_ci.tf
@@ -5,7 +5,7 @@ module "v4_arc_cluster" {
   cpu_nodepool_name = "cpu-nodepool"
   cpu_node_count    = 1
   tpu_nodepool_name = "tpu-nodepool"
-  max_tpu_nodes     = 1
+  max_tpu_nodes     = 2
   github_repo_url   = "https://github.com/pytorch/xla"
   # Dockerfile for this image can be found at test/tpu/Dockerfile
   runner_image      = "gcr.io/tpu-pytorch/tpu-ci-runner:latest"