关于分布式插件pytest-xdist进程级并发参数化说明
UI自动化脚本耗费时间较长，效率低下，我们该如何处理这种情况，提升测试效率，下面我会就分布式插件pytest-xdist的一个动态参数化问题，简单说明一下pytest-xdist的工作流程，以及数据收集过程，希望对大家有所帮助。
–成都-阿木木
说明：
场景：我有100个UI自动化用例，假设每个用例执行时间为一分钟，那么我顺序执行需要执行100分钟，在敏捷开发模式下，耗时比较长，如果赶上项目上线，需要快速得到项目上线后的反馈，那么效率是非常低下的。使用pytest-xdist可以指定多个worker，进行执行用例，我指定10个worker，那么我整个测试脚本执行下来的理论时间为10分钟（PS:只是理论时间），大大的提升了效率。
分布式用例设计原则
- 用例独立，独立运行
- 用例没有顺序，随机运行
- 用例重复运行，运行结束即清理
- 用例的参数化使用非动态数据（重点：下面讲）
pytest-xdist安装
pip install pytest-xdist
pytest-xdist使用
在pytest的命令行参数中指定-n x(运行的进程数量)
-
-n auto:可以自动获取系统的CPU核数，启用该参数CPU占用率会非常高，每个进程执行速度会非常慢，所以，worker越多，并不会按照理论时间来进行测试脚本运行效率的提升 -
-n x:手动指定CPU数量
pytest-xdist分布式执行用例参数化说明
这是一个测试路由器的case，使用了pytest.mark.parametrize进行参数生成，它的作用和ddt数据驱动框架一样，会执行两次测试用例，第一次传递参数“”，第二次传递一个随机10位的字符串参数
@allure.story("Service")
@allure.severity("normal")
@allure.description("服务模块-路由器列表-新建路由器-序列号")
@allure.testcase("XXX", name="测试用例位置")
@allure.title("服务模块-路由器列表-新建路由器-序列号")
@pytest.mark.parametrize("router_sn", ("", PubMethod.random_string("bnuiowehosdc235342", 10)))
def test_617(self, login_page_class, service_page_class, function_driver, router_sn):
logging.info("进入测试")
login_page_class.mode_login_into_url()
service_page_class.click_free_alert_close_icon()
service_page_class.click_create_router_btn()
service_page_class.send_keys_router_name(PubMethod.random_string("hfbudsio2353256", 10))
service_page_class.send_keys_router_sn(router_sn)
service_page_class.click_router_alert_determine_btn()
tips_by_router_sn = service_page_class.get_tips_by_router_sn()
if len(router_sn) == 0:
AssertMethod.assert_equal_screen_shot(function_driver, (tips_by_router_sn, "请输入路由器序列号!"))
else:
AssertMethod.assert_equal_screen_shot(function_driver, (tips_by_router_sn, "请输入正确的路由器序列号!"))
这是pytest的命令行启动参数，我指定了-n参数为auto，运行run.py，框架路径：https://github.com/chineseluo/ui_auto_frame_v2，可以自己下载测试
test_args = ['-s', '-q', '-n=auto', '--browser={}'.format(browser), '--browser_opt={}'.format(browser_opt),
'--type_driver={}'.format(type_driver)]
bringing up nodes...
[gw0] Python 3.8.2 (tags/v3.8.2:7b3ab59, Feb 25 2020, 23:03:10) [MSC v.1916 64 bit (AMD64)]
[gw1] Python 3.8.2 (tags/v3.8.2:7b3ab59, Feb 25 2020, 23:03:10) [MSC v.1916 64 bit (AMD64)]
[gw2] Python 3.8.2 (tags/v3.8.2:7b3ab59, Feb 25 2020, 23:03:10) [MSC v.1916 64 bit (AMD64)]
[gw3] Python 3.8.2 (tags/v3.8.2:7b3ab59, Feb 25 2020, 23:03:10) [MSC v.1916 64 bit (AMD64)]
[gw4] Python 3.8.2 (tags/v3.8.2:7b3ab59, Feb 25 2020, 23:03:10) [MSC v.1916 64 bit (AMD64)]
[gw5] Python 3.8.2 (tags/v3.8.2:7b3ab59, Feb 25 2020, 23:03:10) [MSC v.1916 64 bit (AMD64)]
[gw6] Python 3.8.2 (tags/v3.8.2:7b3ab59, Feb 25 2020, 23:03:10) [MSC v.1916 64 bit (AMD64)]
[gw7] Python 3.8.2 (tags/v3.8.2:7b3ab59, Feb 25 2020, 23:03:10) [MSC v.1916 64 bit (AMD64)]
gw0 [41] / gw1 [41] / gw2 [41] / gw3 [41] / gw4 [41] / gw5 [41] / gw6 [41] / gw7 [41]
scheduling tests via LoadScheduling
=================================== ERRORS ====================================
____________________________ ERROR collecting gw1 _____________________________
Different tests were collected between gw0 and gw1. The difference is:
--- gw0
+++ gw1
@@ -22,7 +22,7 @@
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_615
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_616
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_617[]
-TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_617[45u333isnn]
+TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_617[iw543ui332]
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_618_1
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_618_2
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_618_3
____________________________ ERROR collecting gw2 _____________________________
Different tests were collected between gw0 and gw2. The difference is:
--- gw0
+++ gw2
@@ -22,7 +22,7 @@
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_615
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_616
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_617[]
-TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_617[45u333isnn]
+TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_617[is4no2hw55]
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_618_1
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_618_2
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_618_3
____________________________ ERROR collecting gw3 _____________________________
Different tests were collected between gw0 and gw3. The difference is:
--- gw0
+++ gw3
可以看到我一用启用了8个worker，gw0~gw7，读取的系统默认CPU核数为8，gw0~gw1、gw0~gw2、gw0~gw3…以及其他几个worker报错：
Different tests were collected between gw0 and gw1. The difference is:--- gw0 +++ gw1
可以看到在test_617的两条case异常，有一个+号和减号，分析xdist用例收集，进程调度原理：
收集caseID及进程分发说明：根据简单的示意图可以看出，每一个worker（gwxxx)都会去进行一次标准caseID的收集，整个caseID的收集过程和pytest本身收集caseID的过程是一致的，每一个worker收集完caseID后发送给主节点（管理节点），主节点进行caseID的检查，包括了测试数据的检查，保证每个worker获取的case的数据是一致的，"generating tests twice must produce identical tests" 。在所有worker的测试用例集合相同的前提下，进行测试任务的分发，主节点下发的不再是caseID，而是测试case的索引，以方便告诉每个worker执行哪一个测试用例。
那么现在可以知道该异常出现的原因了，也就是说pytest-xdist在进行caseID收集的时候发现，参数化的case的入参不一致，我们是随机生成的，它只支持静态的有序的入参，下面我们修改一下脚本进行验证。
@allure.story("Service")
@allure.severity("normal")
@allure.description("服务模块-路由器列表-新建路由器-序列号")
@allure.testcase("XXX", name="测试用例位置")
@allure.title("服务模块-路由器列表-新建路由器-序列号")
def test_617_1(self, login_page_class, service_page_class, function_driver):
logging.info("进入测试")
login_page_class.mode_login_into_url()
service_page_class.click_free_alert_close_icon()
service_page_class.click_create_router_btn()
service_page_class.send_keys_router_name(PubMethod.random_string("hfbudsio2353256", 10))
service_page_class.send_keys_router_sn("")
service_page_class.click_router_alert_determine_btn()
tips_by_router_sn = service_page_class.get_tips_by_router_sn()
AssertMethod.assert_equal_screen_shot(function_driver, (tips_by_router_sn, "请输入路由器序列号!"))
@allure.story("Service")
@allure.severity("normal")
@allure.description("服务模块-路由器列表-新建路由器-序列号")
@allure.testcase("XXX", name="测试用例位置")
@allure.title("服务模块-路由器列表-新建路由器-序列号")
def test_617_2(self, login_page_class, service_page_class, function_driver):
logging.info("进入测试")
login_page_class.mode_login_into_url()
service_page_class.click_free_alert_close_icon()
service_page_class.click_create_router_btn()
service_page_class.send_keys_router_name(PubMethod.random_string("hfbudsio2353256", 10))
service_page_class.send_keys_router_sn(PubMethod.random_string("bnuiowehosdc235342", 10))
service_page_class.click_router_alert_determine_btn()
AssertMethod.assert_equal_screen_shot(function_driver, (tips_by_router_sn, "请输入正确的路由器序列号!"))
检查日志，查看caseID的收集是否异常，可以看到下面的日志中，用例的ID收集无异常，验证通过。
bringing up nodes...
[gw0] Python 3.8.2 (tags/v3.8.2:7b3ab59, Feb 25 2020, 23:03:10) [MSC v.1916 64 bit (AMD64)]
[gw1] Python 3.8.2 (tags/v3.8.2:7b3ab59, Feb 25 2020, 23:03:10) [MSC v.1916 64 bit (AMD64)]
[gw2] Python 3.8.2 (tags/v3.8.2:7b3ab59, Feb 25 2020, 23:03:10) [MSC v.1916 64 bit (AMD64)]
[gw3] Python 3.8.2 (tags/v3.8.2:7b3ab59, Feb 25 2020, 23:03:10) [MSC v.1916 64 bit (AMD64)]
[gw4] Python 3.8.2 (tags/v3.8.2:7b3ab59, Feb 25 2020, 23:03:10) [MSC v.1916 64 bit (AMD64)]
[gw5] Python 3.8.2 (tags/v3.8.2:7b3ab59, Feb 25 2020, 23:03:10) [MSC v.1916 64 bit (AMD64)]
[gw6] Python 3.8.2 (tags/v3.8.2:7b3ab59, Feb 25 2020, 23:03:10) [MSC v.1916 64 bit (AMD64)]
[gw7] Python 3.8.2 (tags/v3.8.2:7b3ab59, Feb 25 2020, 23:03:10) [MSC v.1916 64 bit (AMD64)]
gw0 [41] / gw1 [41] / gw2 [41] / gw3 [41] / gw4 [41] / gw5 [41] / gw6 [41] / gw7 [41]
scheduling tests via LoadScheduling
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_606_1
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_605_1_and_2[普通用户]
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_606_2
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_606_4
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_606_3
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_606_5
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_605_1_and_2[管理员]
TestCases/Service/test_servicePageCase.py::TestServicePageCase::test_606_6
总结：
通过分析pytest-xdist的参数化异常原因，可以了解到pytest-xdist的工作流程原理，针对性的处理异常，如果需要使用到pytest-xdist插件的小伙伴，可以看一下，做一下测试，进行简单验证加深了解。
欢迎加入测试交流群：夜行者自动化测试（816489363）进行交流学习QAQ
7053
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
