From 81f3c09d6d46917052fc62759a93bd5a797968ef Mon Sep 17 00:00:00 2001 From: zhangxinyuehfad <59153331+zhangxinyuehfad@users.noreply.github.com> Date: Thu, 5 Feb 2026 23:43:57 +0800 Subject: [PATCH] [CI] Change A2 runner (#6557) ### What this PR does / why we need it? This PR updates the CI runner from `linux-aarch64-a2-*` to `linux-aarch64-a2b3-*` in various test configuration files. This change is necessary to adapt to updates in the CI infrastructure. ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? The changes are configuration updates for CI tests. The correctness will be verified by the CI pipeline. Signed-off-by: hfadzxy --- .../_e2e_nightly_single_node_models.yaml | 2 +- .github/workflows/labled_doctest.yaml | 2 +- .github/workflows/labled_download_model.yaml | 2 +- .github/workflows/pr_test_full.yaml | 4 ++-- .github/workflows/pr_test_light.yaml | 4 ++-- .../workflows/schedule_nightly_test_a2.yaml | 20 +++++++++---------- .../workflows/schedule_test_vllm_main.yaml | 2 +- .../e2e/models/configs/InternVL3_5-8B-hf.yaml | 1 - .../single_node/models/test_qwen3_32b_int8.py | 4 ++-- .../single_node/models/test_qwen3_next.py | 4 ++-- 10 files changed, 22 insertions(+), 23 deletions(-) diff --git a/.github/workflows/_e2e_nightly_single_node_models.yaml b/.github/workflows/_e2e_nightly_single_node_models.yaml index 0064d2e8..82b8e28f 100644 --- a/.github/workflows/_e2e_nightly_single_node_models.yaml +++ b/.github/workflows/_e2e_nightly_single_node_models.yaml @@ -108,7 +108,7 @@ jobs: pip install -v -e . - name: Install tensorflow (for Molmo-7B-D-0924) - if: ${{ inputs.runner == 'linux-aarch64-a2-1' && contains(inputs.model_list, 'Molmo-7B-D-0924') }} + if: ${{ inputs.runner == 'linux-aarch64-a2b3-1' && contains(inputs.model_list, 'Molmo-7B-D-0924') }} shell: bash -l {0} run: | pip install tensorflow==2.19.1 --no-cache-dir diff --git a/.github/workflows/labled_doctest.yaml b/.github/workflows/labled_doctest.yaml index fc822559..cdfed7ea 100644 --- a/.github/workflows/labled_doctest.yaml +++ b/.github/workflows/labled_doctest.yaml @@ -46,7 +46,7 @@ jobs: matrix: vllm_verison: [releases-v0.13.0, releases-v0.13.0-openeuler, main, main-openeuler] name: vLLM Ascend test - runs-on: linux-aarch64-a2-1 + runs-on: linux-aarch64-a2b3-1 container: image: swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/vllm-ascend:${{ matrix.vllm_verison }} steps: diff --git a/.github/workflows/labled_download_model.yaml b/.github/workflows/labled_download_model.yaml index 2a5ccedc..7f50536f 100644 --- a/.github/workflows/labled_download_model.yaml +++ b/.github/workflows/labled_download_model.yaml @@ -19,7 +19,7 @@ jobs: download-models: if: contains(github.event.pull_request.labels.*.name, 'model-download') name: Download models from ModelScope - runs-on: linux-aarch64-a2-0 + runs-on: linux-aarch64-a2b3-0 container: image: swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/vllm-ascend:nightly-cpu diff --git a/.github/workflows/pr_test_full.yaml b/.github/workflows/pr_test_full.yaml index 5bd82e18..1b7e0ade 100644 --- a/.github/workflows/pr_test_full.yaml +++ b/.github/workflows/pr_test_full.yaml @@ -39,7 +39,7 @@ concurrency: jobs: changes: - runs-on: linux-aarch64-a2-0 + runs-on: linux-aarch64-a2b3-0 if: ${{ contains(github.event.pull_request.labels.*.name, 'ready') && contains(github.event.pull_request.labels.*.name, 'ready-for-test') }} outputs: e2e_tracker: ${{ steps.filter.outputs.e2e_tracker }} @@ -81,7 +81,7 @@ jobs: uses: ./.github/workflows/_e2e_test.yaml with: vllm: ${{ matrix.vllm_version }} - runner: linux-aarch64-a2 + runner: linux-aarch64-a2b3 image: swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/cann:8.5.0-910b-ubuntu22.04-py3.11 contains_310: false type: full diff --git a/.github/workflows/pr_test_light.yaml b/.github/workflows/pr_test_light.yaml index 4f8086b8..108e0be9 100644 --- a/.github/workflows/pr_test_light.yaml +++ b/.github/workflows/pr_test_light.yaml @@ -43,7 +43,7 @@ jobs: with: vllm: d7e17aaacd5ed1b4b4be6bcfef3a1b7cbc84fc9a changes: - runs-on: linux-aarch64-a2-0 + runs-on: linux-aarch64-a2b3-0 outputs: e2e_tracker: ${{ steps.filter.outputs.e2e_tracker }} ut_tracker: ${{ steps.filter.outputs.ut_tracker }} @@ -107,7 +107,7 @@ jobs: uses: ./.github/workflows/_e2e_test.yaml with: vllm: ${{ matrix.vllm_version }} - runner: linux-aarch64-a2 + runner: linux-aarch64-a2b3 image: swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/cann:8.5.0-910b-ubuntu22.04-py3.11 contains_310: ${{ needs.changes.outputs._310_tracker == 'true' }} type: light diff --git a/.github/workflows/schedule_nightly_test_a2.yaml b/.github/workflows/schedule_nightly_test_a2.yaml index c24e6060..6bca7a90 100644 --- a/.github/workflows/schedule_nightly_test_a2.yaml +++ b/.github/workflows/schedule_nightly_test_a2.yaml @@ -50,19 +50,19 @@ jobs: matrix: test_config: - name: qwen3-next - os: linux-aarch64-a2-4 + os: linux-aarch64-a2b3-4 tests: tests/e2e/nightly/single_node/models/test_qwen3_next.py - name: qwen3-32b - os: linux-aarch64-a2-4 + os: linux-aarch64-a2b3-4 tests: tests/e2e/nightly/single_node/models/test_qwen3_32b.py - name: qwen3-32b-in8-a2 - os: linux-aarch64-a2-4 + os: linux-aarch64-a2b3-4 tests: tests/e2e/nightly/single_node/models/test_qwen3_32b_int8.py - name: test_custom_op - os: linux-aarch64-a2-1 + os: linux-aarch64-a2b3-1 tests: tests/e2e/nightly/single_node/ops/singlecard_ops - name: test_custom_op_multi_card - os: linux-aarch64-a2-4 + os: linux-aarch64-a2b3-4 tests: tests/e2e/nightly/single_node/ops/multicard_ops_a2/ uses: ./.github/workflows/_e2e_nightly_single_node.yaml with: @@ -108,26 +108,26 @@ jobs: fail-fast: false matrix: test_config: - - os: linux-aarch64-a2-1 + - os: linux-aarch64-a2b3-1 model_list: - Qwen3-8B - Qwen2-Audio-7B-Instruct - Qwen3-8B-W8A8 - Qwen3-VL-8B-Instruct - Qwen2.5-Omni-7B - - os: linux-aarch64-a2-1 + - os: linux-aarch64-a2b3-1 model_list: - ERNIE-4.5-21B-A3B-PT - InternVL3_5-8B-hf - Molmo-7B-D-0924 - Llama-3.2-3B-Instruct - llava-onevision-qwen2-0.5b-ov-hf - - os: linux-aarch64-a2-2 + - os: linux-aarch64-a2b3-2 model_list: - Qwen3-30B-A3B - Qwen3-VL-30B-A3B-Instruct - Qwen3-30B-A3B-W8A8 - - os: linux-aarch64-a2-4 + - os: linux-aarch64-a2b3-4 model_list: - Qwen3-Next-80B-A3B-Instruct - Qwen3-Omni-30B-A3B-Instruct @@ -147,7 +147,7 @@ jobs: fail-fast: false matrix: vllm_verison: [releases-v0.13.0, releases-v0.13.0-openeuler, main, main-openeuler] - runs-on: linux-aarch64-a2-1 + runs-on: linux-aarch64-a2b3-1 container: image: swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/vllm-ascend:${{ matrix.vllm_verison }} steps: diff --git a/.github/workflows/schedule_test_vllm_main.yaml b/.github/workflows/schedule_test_vllm_main.yaml index 3a73af74..6e736a4e 100644 --- a/.github/workflows/schedule_test_vllm_main.yaml +++ b/.github/workflows/schedule_test_vllm_main.yaml @@ -34,7 +34,7 @@ jobs: uses: ./.github/workflows/_e2e_test.yaml with: vllm: main - runner: linux-aarch64-a2 + runner: linux-aarch64-a2b3 image: swr.cn-southwest-2.myhuaweicloud.com/base_image/ascend-ci/cann:8.5.0-910b-ubuntu22.04-py3.11 contains_310: false type: full diff --git a/tests/e2e/models/configs/InternVL3_5-8B-hf.yaml b/tests/e2e/models/configs/InternVL3_5-8B-hf.yaml index 3628785a..6aae094b 100644 --- a/tests/e2e/models/configs/InternVL3_5-8B-hf.yaml +++ b/tests/e2e/models/configs/InternVL3_5-8B-hf.yaml @@ -1,5 +1,4 @@ model_name: "OpenGVLab/InternVL3_5-8B-hf" -runner: "linux-aarch64-a2-1" hardware: "Atlas A2 Series" model: "vllm-vlm" tasks: diff --git a/tests/e2e/nightly/single_node/models/test_qwen3_32b_int8.py b/tests/e2e/nightly/single_node/models/test_qwen3_32b_int8.py index a9c7d7a8..f2507028 100644 --- a/tests/e2e/nightly/single_node/models/test_qwen3_32b_int8.py +++ b/tests/e2e/nightly/single_node/models/test_qwen3_32b_int8.py @@ -45,10 +45,10 @@ api_keyword_args = { } batch_size_dict = { - "linux-aarch64-a2-4": 72, + "linux-aarch64-a2b3-4": 72, "linux-aarch64-a3-4": 76, } -VLLM_CI_RUNNER = os.getenv("VLLM_CI_RUNNER", "linux-aarch64-a2-4") +VLLM_CI_RUNNER = os.getenv("VLLM_CI_RUNNER", "linux-aarch64-a2b3-4") performance_batch_size = batch_size_dict.get(VLLM_CI_RUNNER, 1) aisbench_cases = [{ diff --git a/tests/e2e/nightly/single_node/models/test_qwen3_next.py b/tests/e2e/nightly/single_node/models/test_qwen3_next.py index 7d35c8f4..c05fab6a 100644 --- a/tests/e2e/nightly/single_node/models/test_qwen3_next.py +++ b/tests/e2e/nightly/single_node/models/test_qwen3_next.py @@ -27,10 +27,10 @@ api_keyword_args = { } batch_size_dict = { - "linux-aarch64-a2-4": 64, + "linux-aarch64-a2b3-4": 64, "linux-aarch64-a3-4": 64, } -VLLM_CI_RUNNER = os.getenv("VLLM_CI_RUNNER", "linux-aarch64-a2-4") +VLLM_CI_RUNNER = os.getenv("VLLM_CI_RUNNER", "linux-aarch64-a2b3-4") performance_batch_size = batch_size_dict.get(VLLM_CI_RUNNER, 1) aisbench_cases = [{