socket hang up when working with Jenkins and use 'start_process'

[jackniu81]

when working with Jenkins, and use 'start_process' to ask Jenkins to auto start ChromeDriver, but it fails.

Environment

1. Node.js: 10.15.0
2. Chrome Driver: 2.45.615291
3. Windows 2008 R2,
4. nightwatch: 1.0.18

nightwatch config:

module.exports = {
    "src_folders": [
        "tests"
    ],
    "globals_path": "./globalsModule.js",
    "webdriver": {
        "start_process": true,     // here!! Jenkins cannot auto start chromedriver
        "server_path": "node_modules/chromedriver/lib/chromedriver/chromedriver.exe",
        "port": 9515
    },
    "test_settings": {
        "default": {
            "desiredCapabilities": {
                "browserName": "chrome",
                "chromeOptions": {
                    "args": [
                        // use this to use 'headless' mode
                        // "headless", "no-sandbox", "disable-gpu"
                    ]
                },
            }
        }
    },
    "test_workers": {   //  runs the tests in parallel
        "enabled": true,
        "workers": "auto"
    }
};

Error
08:44:09 tests\report.js [Report] Test Suite
08:44:09 tests\report.js ===================
08:44:09 tests\report.js POST /session - ECONNRESET
08:44:09 tests\report.js Error: socket hang up
08:44:09 tests\report.js at createHangUpError (_http_client.js:323:15)
08:44:09 tests\report.js at Socket.socketCloseListener (_http_client.js:364:25)
08:44:09 tests\report.js An occurred error while retrieving a new session: "socket hang up"
08:44:09 tests\report.js at Socket.socketCloseListener (_http_client.js:364:11)

[JamesODonoghue]

I had a bunch of issues in Jenkins after upgrading to the latest version of nightwatch.

What helped was upgrading the chrome binary on the jenkins server, then upgrading chromedriver to support that version of chrome.

We use selenium so I upgraded that as well.

Jenkins server:
Node 10.12.0
Linux Redhat v ?
ChromeDriver 73.0.3683.68
Selenium 3.8.1

No issues in Jenkins now. Not sure if this helps.

[aberonni]

Could be a duplicate of #1936 ?

[beatfactor]

How about this workaround?

I'm going to increase the default timeout.