From c7dbc25902321b1e1fae9caeba5fa6ab957c1460 Mon Sep 17 00:00:00 2001 From: Evan Lezar Date: Tue, 13 May 2025 22:24:37 +0200 Subject: [PATCH 1/2] [no-relnote] Use cuda 12.9.0 image in tests Signed-off-by: Evan Lezar --- tests/e2e/nvidia-container-toolkit_test.go | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/tests/e2e/nvidia-container-toolkit_test.go b/tests/e2e/nvidia-container-toolkit_test.go index 2b1ef289..59abedef 100644 --- a/tests/e2e/nvidia-container-toolkit_test.go +++ b/tests/e2e/nvidia-container-toolkit_test.go @@ -173,10 +173,10 @@ var _ = Describe("docker", Ordered, ContinueOnFailure, func() { When("Testing CUDA Forward compatibility", Ordered, func() { BeforeAll(func(ctx context.Context) { - _, _, err := runner.Run("docker pull nvcr.io/nvidia/cuda:12.8.0-base-ubi8") + _, _, err := runner.Run("docker pull nvcr.io/nvidia/cuda:12.9.0-base-ubi8") Expect(err).ToNot(HaveOccurred()) - compatOutput, _, err := runner.Run("docker run --rm -i -e NVIDIA_VISIBLE_DEVICES=void nvcr.io/nvidia/cuda:12.8.0-base-ubi8 bash -c \"ls /usr/local/cuda/compat/libcuda.*.*\"") + compatOutput, _, err := runner.Run("docker run --rm -i -e NVIDIA_VISIBLE_DEVICES=void nvcr.io/nvidia/cuda:12.9.0-base-ubi8 bash -c \"ls /usr/local/cuda/compat/libcuda.*.*\"") Expect(err).ToNot(HaveOccurred()) Expect(compatOutput).ToNot(BeEmpty()) @@ -199,19 +199,19 @@ var _ = Describe("docker", Ordered, ContinueOnFailure, func() { }) It("should work with the nvidia runtime in legacy mode", func(ctx context.Context) { - ldconfigOut, _, err := runner.Run("docker run --rm -i -e NVIDIA_DISABLE_REQUIRE=true --runtime=nvidia --gpus all nvcr.io/nvidia/cuda:12.8.0-base-ubi8 bash -c \"ldconfig -p | grep libcuda.so.1\"") + ldconfigOut, _, err := runner.Run("docker run --rm -i -e NVIDIA_DISABLE_REQUIRE=true --runtime=nvidia --gpus all nvcr.io/nvidia/cuda:12.9.0-base-ubi8 bash -c \"ldconfig -p | grep libcuda.so.1\"") Expect(err).ToNot(HaveOccurred()) Expect(ldconfigOut).To(ContainSubstring("/usr/local/cuda/compat")) }) It("should work with the nvidia runtime in CDI mode", func(ctx context.Context) { - ldconfigOut, _, err := runner.Run("docker run --rm -i -e NVIDIA_DISABLE_REQUIRE=true --runtime=nvidia -e NVIDIA_VISIBLE_DEVICES=runtime.nvidia.com/gpu=all nvcr.io/nvidia/cuda:12.8.0-base-ubi8 bash -c \"ldconfig -p | grep libcuda.so.1\"") + ldconfigOut, _, err := runner.Run("docker run --rm -i -e NVIDIA_DISABLE_REQUIRE=true --runtime=nvidia -e NVIDIA_VISIBLE_DEVICES=runtime.nvidia.com/gpu=all nvcr.io/nvidia/cuda:12.9.0-base-ubi8 bash -c \"ldconfig -p | grep libcuda.so.1\"") Expect(err).ToNot(HaveOccurred()) Expect(ldconfigOut).To(ContainSubstring("/usr/local/cuda/compat")) }) It("should NOT work with nvidia-container-runtime-hook", func(ctx context.Context) { - ldconfigOut, _, err := runner.Run("docker run --rm -i -e NVIDIA_DISABLE_REQUIRE=true --runtime=runc --gpus all nvcr.io/nvidia/cuda:12.8.0-base-ubi8 bash -c \"ldconfig -p | grep libcuda.so.1\"") + ldconfigOut, _, err := runner.Run("docker run --rm -i -e NVIDIA_DISABLE_REQUIRE=true --runtime=runc --gpus all nvcr.io/nvidia/cuda:12.9.0-base-ubi8 bash -c \"ldconfig -p | grep libcuda.so.1\"") Expect(err).ToNot(HaveOccurred()) Expect(ldconfigOut).To(ContainSubstring("/usr/lib64")) }) From 5217270f18b23c1bd44fb64ca491e5b1f3f18ec0 Mon Sep 17 00:00:00 2001 From: Evan Lezar Date: Tue, 13 May 2025 23:18:01 +0200 Subject: [PATCH 2/2] [no-relnote] Use 550 driver in tests Signed-off-by: Evan Lezar Signed-off-by: Carlos Eduardo Arango Gutierrez --- .github/workflows/e2e.yaml | 2 +- tests/e2e/infra/aws.yaml | 1 + 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/.github/workflows/e2e.yaml b/.github/workflows/e2e.yaml index f0f4a829..7c788a38 100644 --- a/.github/workflows/e2e.yaml +++ b/.github/workflows/e2e.yaml @@ -55,7 +55,7 @@ jobs: go-version: ${{ env.GOLANG_VERSION }} - name: Set up Holodeck - uses: NVIDIA/holodeck@v0.2.9 + uses: NVIDIA/holodeck@v0.2.10 with: aws_access_key_id: ${{ secrets.AWS_ACCESS_KEY_ID }} aws_secret_access_key: ${{ secrets.AWS_SECRET_ACCESS_KEY }} diff --git a/tests/e2e/infra/aws.yaml b/tests/e2e/infra/aws.yaml index d73bf7f5..b512fc78 100644 --- a/tests/e2e/infra/aws.yaml +++ b/tests/e2e/infra/aws.yaml @@ -28,3 +28,4 @@ spec: install: false nvidiaDriver: install: true + branch: 550