我们目前有一个包含24个integration/*.js文件的存储库。我们已经通过TeamCity CI工作流程成功运行了赛普拉斯测试，但是，我无法使这些测试并行运行。
通过配置方式，我们将回购/测试连接到dashboard.cypress.io中的组织。与仪表板键一起，我们在TeamCity构建步骤中通过npm run调用Cypress，并在package.json scripts{}中使用以下配置：
"cypress:run:qa": "cypress run -c cypress.qa.json --record --key xyxyxyx-xyxyx-xyxyx-xyxyx-xyxyxyxyxyxy --parallel --config baseUrl=\"https://some.url.com\" --ci-build-id=%BUILD_ID_CYPRESS%"
上面传递的％BUILD_ID_CYPRESS％环境变量是TeamCity版本ID和编号的自定义组合，以使其唯一。
package.json列出以下依赖项：
"devDependencies": {
"cypress": "^3.3.2",
"cypress-plugin-retries": "^1.2.0"
},
"dependencies": {
"ldclient-electron": "^1.0.1",
"ldclient-js": "^2.10.0",
"ldclient-node": "^5.7.4"
}
cypress.qa.json配置文件的内容是：
{
"projectId": "*redacted*",
"baseUrl": "https://some.url.com",
"numTestsKeptInMemory": 50,
"port": 3002,
"reporter": "teamcity",
"reporterOptions": null,
"watchForFileChanges": true,
"viewportWidth": 1280,
"viewportHeight": 720,
"defaultCommandTimeout": 30000,
"execTimeout": 90000,
"taskTimeout": 90000,
"pageLoadTimeout": 60000,
"requestTimeout": 60000,
"responseTimeout": 60000,
"screenshotsFolder": "cypress/screenshots",
"trashAssetsBeforeRuns": true,
"videoCompression": 32,
"videosFolder": "cypress/videos",
"video": true,
"videoUploadOnPasses": false,
"env": {
"retryAttempts": "2",
"*redacted URL Name*": "https://some.otherURL.com/someSubDirectory/"
}
}
[在TeamCity中，我在构建日志中看到以下输出，这似乎表明赛普拉斯正在正确传递并行配置：
[15:48:08] [Step 8/11] ================================================================================
[15:48:08] [Step 8/11]
[15:48:08] [Step 8/11] (Run Starting)
[15:48:08] [Step 8/11]
[15:48:09] [Step 8/11] ┌────────────────────────────────────────────────────────────────────────────�
[15:48:09] [Step 8/11] │ Cypress: 3.4.0 │
[15:48:09] [Step 8/11] │ Browser: Electron 61 (headless) │
[15:48:09] [Step 8/11] │ Specs: 24 found (**spec names redacted*** │
[15:48:09] [Step 8/11] │ Params: Group: false, Parallel: true │
[15:48:09] [Step 8/11] │ Run URL: https://dashboard.cypress.io/#/projects/*redactedProjID*/runs/26 │
[15:48:09] [Step 8/11] └────────────────────────────────────────────────────────────────────────────┘
[15:48:09] [Step 8/11]
[15:48:09] [Step 8/11] 2019-07-23T15:48:10.502Z cypress:server:record before spec run { spec: undefined }
[15:48:09] [Step 8/11] 2019-07-23T15:48:10.503Z cypress:server:api request to url: POST https://api.cypress.io/runs/*redactedGUID*/instances with params: {"body":{"spec":null,"groupId":"win32-Electron-61-abcabcabc","machineId":"**RedactedMachineID","platform":{"osCpus":[{"model":"Intel(R) Xeon(R) CPU E5-2676 v3 @ 2.40GHz","speed":2400,"times":{"user":*redacted*,"nice":0,"sys":*redacted*,"idle":*redacted*,"irq":*redacted*}},{"model":"Intel(R) Xeon(R) CPU E5-2676 v3 @ 2.40GHz","speed":2400,"times":{"user":*redacted*,"nice":0,"sys":*redacted*,"idle":8866723484,"irq":517593}}],"osName":"win32","osMemory":{"free":956882944,"total":4294557696},"osVersion":"6.3.9600","browserName":"Electron","browserVersion":"61.0.3163.100"}},"headers":{"x-route-version":"5","x-os-name":"win32","x-cypress-version":"3.4.0"}} and token: undefined
[15:48:09] [Step 8/11] 2019-07-23T15:48:10.624Z cypress:server:api response { spec: 'cypress\\integration\\*redacted*.js', instanceId: '*redactedGUID*', claimedInstances: 1, totalInstances: 24, estimatedWallClockDuration: 675698 }
但是，每次后续测试运行的日志输出都会重复相同的machineId，所有这些测试都在同一TeamCity代理上串行运行。
我们对赛普拉斯中的测试分组并没有做任何花哨的事情，也没有看到任何其他特别独特的东西，但是赛普拉斯的运行继续按顺序执行。
我特别担心其中之一：
我不确定您的方法。但是其他选项很少。
一种方法是在TC上使用docker-compose运行测试。这是一个示例项目。该脚本将为您的项目创建一个docker映像，并运行该映像的多个实例。
https://github.com/SgSridhar/cypress-docker-parallel-tests/blob/master/package.json#L25
我已经对此加阵了一段时间。赛普拉斯github动作似乎是以您想要的任何模式运行测试的最佳选择。
我建议采取github行动方法。您可以节省大量时间。