Update TPU ARC cluster (#7677)

pytorch · Jul 12, 2024 · 8390de2 · 8390de2
1 parent 1651e76
commit 8390de2
Show file tree

Hide file tree

Showing 4 changed files with 46 additions and 5 deletions.
diff --git a/.gitignore b/.gitignore
@@ -23,7 +23,7 @@ torch_xla/csrc/version.cpp
 /docs/src/*
 
 # Local terraform state
-.terraform*
+.terraform
 
 
 # Build system temporary files

diff --git a/infra/terraform_modules/arc_v4_container_cluster/main.tf b/infra/terraform_modules/arc_v4_container_cluster/main.tf
@@ -73,7 +73,7 @@ resource "google_container_node_pool" "arc_v4_tpu_nodes" {
 
 resource "helm_release" "arc" {
   name             = "actions-runner-controller"
-  chart            = "oci://ghcr.io/actions/actions-runner-controller-charts/gha-runner-scale-set-controller"
+  chart            = "oci://ghcr.io/actions/actions-runner-controller-charts/gha-runner-scale-set-controller:0.9.3"
   namespace        = var.arc_namespace
   create_namespace = true
 }
@@ -83,7 +83,7 @@ resource "helm_release" "arc_runner_set" {
   depends_on = [
     helm_release.arc
   ]
-  chart            = "oci://ghcr.io/actions/actions-runner-controller-charts/gha-runner-scale-set"
+  chart            = "oci://ghcr.io/actions/actions-runner-controller-charts/gha-runner-scale-set:0.9.3"
   namespace        = var.runner_namespace
   create_namespace = true
 

diff --git a/infra/tpu-pytorch/.terraform.lock.hcl b/infra/tpu-pytorch/.terraform.lock.hcl
diff --git a/test/tpu/Dockerfile b/test/tpu/Dockerfile
@@ -3,12 +3,12 @@ FROM us-central1-docker.pkg.dev/tpu-pytorch-releases/docker/development:tpu as b
 # Replace value with the latest runner release version
 # source: https://github.com/actions/runner/releases
 # ex: 2.303.0
-ARG RUNNER_VERSION="2.316.1"
+ARG RUNNER_VERSION="2.317.0"
 ARG RUNNER_ARCH="x64"
 # Replace value with the latest runner-container-hooks release version
 # source: https://github.com/actions/runner-container-hooks/releases
 # ex: 0.3.1
-ARG RUNNER_CONTAINER_HOOKS_VERSION="0.6.0"
+ARG RUNNER_CONTAINER_HOOKS_VERSION="0.6.1"
 
 ARG USER=runner