[XPU] [CI] Fix xpu ci bug (#7014)

* fix xpu ci bug

* Remove unnecessary blank line in conftest.py

* Update upload-artifact action to version 6

* Update _xpu_8cards_case_test.yml

* fix ci bug

* Change exit code on test failure to 1

* fix ci bug

* fix ci bug

* fix ci bug

* fix ci bug

* Update conftest.py

.github/workflows/_xpu_4cards_case_test.yml

@@ -193,13 +193,29 @@ jobs:
           echo "============================开始运行pytest测试============================"
           export PYTHONPATH=/workspace/FastDeploy/
           export PYTHONPATH=$(pwd)/tests/xpu_ci:$PYTHONPATH
+          mkdir -p case_logs
+          set +e
           python -m pytest -v -s --tb=short tests/xpu_ci/4cards_cases/
           exit_code=$?
+          set -e
+
+          # 修改case_logs权限，确保Docker外部的runner用户可以读取并上传
+          chmod -R a+rX case_logs/ 2>/dev/null || true
 
           if [ $exit_code -eq 0 ]; then
               echo "============================4卡cases测试通过!============================"
+              exit $exit_code
           else
               echo "============================4卡cases测试失败,请检查日志!============================"
               exit $exit_code
           fi
           '
+
+      - name: Upload case logs
+        if: always()
+        uses: actions/upload-artifact@v6
+        with:
+          name: xpu-4cards-case-logs
+          path: FastDeploy/case_logs/
+          retention-days: 7
+          if-no-files-found: ignore

.github/workflows/_xpu_8cards_case_test.yml

@@ -182,8 +182,14 @@ jobs:
           echo "============================开始运行pytest测试============================"
           export PYTHONPATH=/workspace/FastDeploy/
           export PYTHONPATH=$(pwd)/tests/xpu_ci:$PYTHONPATH
+          mkdir -p case_logs
+          set +e
           python -m pytest -v -s --tb=short tests/xpu_ci/8cards_cases/
           exit_code=$?
+          set -e
+
+          # 修改case_logs权限，确保Docker外部的runner用户可以读取并上传
+          chmod -R a+rX case_logs/ 2>/dev/null || true
 
           if [ $exit_code -eq 0 ]; then
               echo "============================8卡cases测试通过!============================"
@@ -192,3 +198,12 @@ jobs:
               exit $exit_code
           fi
           '
+
+      - name: Upload case logs
+        if: always()
+        uses: actions/upload-artifact@v6
+        with:
+          name: xpu-8cards-case-logs
+          path: FastDeploy/case_logs/
+          retention-days: 7
+          if-no-files-found: ignore

tests/xpu_ci/8cards_cases/test_pd_21b_ep4tp1.py

@@ -109,7 +109,7 @@ def print_pd_logs_on_failure():
     log_dirs = ["log_router", "log_prefill", "log_decode"]
 
     for log_dir in log_dirs:
-        nohup_path = os.path.join(log_dir, "log_0/worklog.0")
+        nohup_path = os.path.join(log_dir, "log_0/workerlog.0")
         if os.path.exists(nohup_path):
             print(f"\n========== {nohup_path} ==========")
             with open(nohup_path, "r") as f:

tests/xpu_ci/8cards_cases/test_pd_21b_ep4tp4.py

@@ -109,7 +109,7 @@ def print_pd_logs_on_failure():
     log_dirs = ["log_router", "log_prefill", "log_decode"]
 
     for log_dir in log_dirs:
-        nohup_path = os.path.join(log_dir, "log_0/worklog.0")
+        nohup_path = os.path.join(log_dir, "log_0/workerlog.0")
         if os.path.exists(nohup_path):
             print(f"\n========== {nohup_path} ==========")
             with open(nohup_path, "r") as f:

tests/xpu_ci/8cards_cases/test_pd_21b_ep4tp4_cudagraph.py

@@ -109,7 +109,7 @@ def print_pd_logs_on_failure():
     log_dirs = ["log_router", "log_prefill", "log_decode"]
 
     for log_dir in log_dirs:
-        nohup_path = os.path.join(log_dir, "log_0/worklog.0")
+        nohup_path = os.path.join(log_dir, "log_0/workerlog.0")
         if os.path.exists(nohup_path):
             print(f"\n========== {nohup_path} ==========")
             with open(nohup_path, "r") as f:

tests/xpu_ci/8cards_cases/test_pd_p_tp4ep4_d_tp1ep4.py

@@ -110,7 +110,7 @@ def print_pd_logs_on_failure():
     log_dirs = ["log_router", "log_prefill", "log_decode"]
 
     for log_dir in log_dirs:
-        nohup_path = os.path.join(log_dir, "log_0/worklog.0")
+        nohup_path = os.path.join(log_dir, "log_0/workerlog.0")
         if os.path.exists(nohup_path):
             print(f"\n========== {nohup_path} ==========")
             with open(nohup_path, "r") as f:

tests/xpu_ci/conftest.py

@@ -23,6 +23,7 @@ XPU CI测试框架 - 通用配置和辅助函数
 4. 环境配置 - 设置XPU相关环境变量
 """
 
+import glob
 import json
 import os
 import shutil
@@ -31,6 +32,8 @@ import time
 
 import pytest
 
+CASE_LOGS_DIR = os.path.join(os.getcwd(), "case_logs")
+
 
 def get_xpu_id():
     """获取XPU_ID环境变量"""
@@ -457,3 +460,42 @@ def setup_logprobs_zmq_env():
         os.environ[key] = value
         print(f"设置环境变量: {key}={value}")
     return original_values
+
+
+# ============ 日志归档 pytest hook ============
+
+
+def _archive_case_logs(test_name):
+    """
+    将当前工作目录下所有 log 开头的文件夹和 server.log 复制到 case_logs/{test_name}/ 下
+    """
+    dest_dir = os.path.join(CASE_LOGS_DIR, test_name)
+    os.makedirs(dest_dir, exist_ok=True)
+
+    # 复制所有 log* 目录
+    for entry in glob.glob("log*"):
+        if os.path.isdir(entry):
+            shutil.copytree(entry, os.path.join(dest_dir, entry), dirs_exist_ok=True)
+        elif os.path.isfile(entry):
+            # 处理 server.log 等 log 开头的文件
+            shutil.copy2(entry, os.path.join(dest_dir, entry))
+
+    # 单独处理 server.log（不以 log 开头但也是关键日志）
+    if os.path.exists("server.log") and not os.path.exists(os.path.join(dest_dir, "server.log")):
+        shutil.copy2("server.log", os.path.join(dest_dir, "server.log"))
+
+
+@pytest.hookimpl(hookwrapper=True, trylast=True)
+def pytest_runtest_makereport(item, call):
+    """每个测试阶段结束后归档日志（仅在 call 阶段后执行）"""
+    outcome = yield
+    report = outcome.get_result()
+
+    if report.when == "call":
+        # 使用测试文件名（不含 .py）作为归档目录名
+        test_file = os.path.basename(item.fspath)
+        test_name = os.path.splitext(test_file)[0]
+        try:
+            _archive_case_logs(test_name)
+        except Exception:
+            pass