[CI] Use CI pool (#428)

Use CI pool instead of self-host for e2e test to speed up CI.

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>

.github/workflows/vllm_ascend_test_main.yaml

@@ -25,7 +25,7 @@ on:
     paths:
       - '*.txt'
       - '**/*.py'
-      - '.github/workflows/vllm_ascend_test.yaml'
+      - '.github/workflows/vllm_ascend_test_main.yaml'
       - '!docs/**'
       - 'pytest.ini'
 
@@ -37,69 +37,30 @@ defaults:
     shell: bash -el {0}
 
 jobs:
-  dispatch:
-    name: vLLM Ascend test (dispatch)
-    runs-on: ascend-ci-arm64
-    outputs:
-      number: ${{ steps.dispatch-device.outputs.number }}
-    steps:
-      - name: vLLM Ascend test (dispatch)
-        id: dispatch-device
-        run: |
-          # Try to acquire lock to dispatch devices
-          lockfile /tmp/dispatch.lock
-
-          # Print npu info
-          npu-list /dev/null 2>&1
-
-          # Select first available device (Skip reserved davinci1 and davinci0)
-          NUMBER=$(npu-list /dev/null 2>&1 | grep None | grep -v davinci1 | grep -v davinci0 | head -1 | cut -b 15)
-          echo "Dispatch to /dev/davinci$NUMBER"
-          echo "number=$NUMBER" >> $GITHUB_OUTPUT
-
   test:
-    needs: [dispatch]
     name: vLLM Ascend test (self-host)
-    runs-on: ascend-ci-arm64  # actionlint-ignore: runner-label
-
+    runs-on: linux-arm64-npu-1  # actionlint-ignore: runner-label
     container:
       image: quay.io/ascend/cann:8.0.0-910b-ubuntu22.04-py3.10
-      volumes:
-        - /usr/local/dcmi:/usr/local/dcmi
-        - /usr/local/bin/npu-smi:/usr/local/bin/npu-smi
-        - /usr/local/Ascend/driver/lib64/:/usr/local/Ascend/driver/lib64/
-        # Use self-host cache speed up pip and model download
-        - /home/action/cache:/github/home/.cache/
-        # for dispatch lock
-        - /tmp/:/tmp/
-        # for vllm and vllm-ascend
-        - /data1/code:/code
-      options: >-
-        --device /dev/davinci${{ needs.dispatch.outputs.number }}
-        --device /dev/davinci_manager
-        --device /dev/devmm_svm
-        --device /dev/hisi_hdc
-      env:
-        HF_ENDPOINT: https://hf-mirror.com
     steps:
       - name: Check npu and CANN info
         run: |
           npu-smi info
           cat /usr/local/Ascend/ascend-toolkit/latest/"$(uname -i)"-linux/ascend_toolkit_install.info
-          # unlock
-          rm -rf /tmp/dispatch.lock
 
       - name: Config mirrors
         run: |
           sed -i 's|ports.ubuntu.com|mirrors.tuna.tsinghua.edu.cn|g' /etc/apt/sources.list
           pip config set global.index-url https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simple
+          apt-get update -y
+          apt install git -y
+          git config --global url."https://gh-proxy.test.osinfra.cn/https://github.com/".insteadOf https://github.com/
 
       - name: Checkout vllm-project/vllm-ascend repo
         uses: actions/checkout@v4
 
       - name: Install system dependencies
         run: |
-          apt-get update -y
           apt-get -y install `cat packages.txt`
 
       - name: Install dependencies
@@ -123,32 +84,38 @@ jobs:
 
       - name: Install pta
         run: |
-          cd /code/pta/
-          pip install ./torch_npu-2.5.1.dev20250320-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
+          if [ ! -d /root/.cache/pta ]; then
+            mkdir -p /root/.cache/pta
+          fi
+
+          if [ ! -f /root/.cache/pta/torch_npu-2.5.1.dev20250320-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl ]; then
+            cd /root/.cache/pta
+            rm -rf pytorch_v2.5.1_py310*
+            wget https://pytorch-package.obs.cn-north-4.myhuaweicloud.com/pta/Daily/v2.5.1/20250320.3/pytorch_v2.5.1_py310.tar.gz
+            tar -zxvf pytorch_v2.5.1_py310.tar.gz
+          fi
+
+          pip install /root/.cache/pta/torch_npu-2.5.1.dev20250320-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
 
       - name: Run vllm-project/vllm-ascend test for V0 Engine
+        env:
+          VLLM_USE_V1: 0
+          HF_ENDPOINT: https://hf-mirror.com
         run: |
           VLLM_USE_V1=0 pytest -sv tests
 
       - name: Run vllm-project/vllm-ascend test for V1 Engine
+        env:
+          VLLM_USE_V1: 1
+          VLLM_WORKER_MULTIPROC_METHOD: spawn
+          HF_ENDPOINT: https://hf-mirror.com
         run: |
-          VLLM_USE_V1=1 VLLM_WORKER_MULTIPROC_METHOD=spawn pytest -sv tests
+          pytest -sv tests
 
       - name: Run vllm-project/vllm test for V0 Engine
         env:
           VLLM_USE_V1: 0
           PYTORCH_NPU_ALLOC_CONF: max_split_size_mb:256
+          HF_ENDPOINT: https://hf-mirror.com
         run: |
           pytest -sv
-
-  post_cleanup:
-    name: vLLM Ascend test (post-cleanup)
-    needs: [test]
-    runs-on: ascend-ci-arm64  # actionlint-ignore: runner-label
-    if: always()
-    steps:
-      - name: Remove dispatch lock if exists
-        run: |
-          if [ -f "/tmp/dispatch.lock" ]; then
-            rm -f "/tmp/dispatch.lock"
-          fi