Getting random error from WebDriver - Connection reset by peer

[mjaeggi]

Bug report

I'm getting the next WebDriver error randomly in different parts of the tests:

WebDriverError: java.net.SocketException: Connection reset by peer (connect failed)

Doesn't matter the action against the browser.

• Node Version: 8.9.4
• Protractor Version: 5.3
• Browser(s): Chrome
• Selenium: 3.11
• ChromeDriver: 2.36
• Operating System and Version: Mac OS - Sierra 10.12.6
• Your protractor configuration file

const chai = require('chai');
const chaiAsPromised = require('chai-as-promised');

const frameworkPath = require.resolve('protractor-cucumber-framework');

module.exports = {
  allScriptsTimeout: 50000,
  seleniumAddress: 'http://localhost:4444/wd/hub',
  baseUrl: '[...]',

  resultJsonOutputFile: 'report.json',
  framework: 'custom',
  frameworkPath,
  capabilities: {
    browserName: 'chrome',
  },
  // How long to wait for a page to load.
  getPageTimeout: 50000,
  onPrepare: () => {
    chai.use(chaiAsPromised);
    // Disable Angular animations so e2e tests run more quickly
    browser.addMockModule('disableNgAnimate', () => {
      angular.module('disableNgAnimate', []).run(($animate) => {
        $animate.enabled(false);
      });
    });

    browser.ignoreSynchronization = true;

    browser.manage().timeouts().implicitlyWait(30000);
    browser.driver.manage().window().setSize(1300, 1000);
  },
  specs: [
    './e2e/features/*/*.feature',
  ],
  SELENIUM_PROMISE_MANAGER: false,
  cucumberOpts: {
    require: './e2e/features/*/*.js',
    format: 'progress-bar',
    keepAlive: false,
  },
};

• Output from running the test

WebDriverError: java.net.SocketException: Connection reset by peer (connect failed)
           at Object.checkLegacyResponse (/Users/.../node_modules/selenium-webdriver/lib/error.js:546:15)
           at parseHttpResponse (/Users/.../node_modules/selenium-webdriver/lib/http.js:509:13)
           at doSend.then.response (/Users/.../node_modules/selenium-webdriver/lib/http.js:441:30)
           at <anonymous>
           at process._tickCallback (internal/process/next_tick.js:188:7)Error
           at ElementArrayFinder.applyAction_ (/Users/.../node_modules/protractor/built/element.js:459:27)
           at ElementArrayFinder.(anonymous function).args [as getAttribute] (/Users/.../node_modules/protractor/built/element.js:91:29)
           at ElementFinder.(anonymous function).args [as getAttribute] (/Users/.../node_modules/protractor/built/element.js:831:22)
           at HomePage.getFontNameFromCardCarousel (/Users/.../tests/qeguitests/e2e/pages/HomePage.js:91:148)
           at World.When (/Users/.../tests/qeguitests/e2e/features/steps_definitions/homepage_steps.js:99:33)
   - And I move to the "right" the carousel # tests/qeguitests/e2e/features/steps_definitions/homepage_steps.js:54

[pawelus]

Duplicate of #4693?
Any getX() operation (eg. getText() or getAttribute()) done on ElementArrayFinder seems to cause this randomly when SELENIUM_PROMISE_MANAGER is set to false.

[mjaeggi]

Yeah, totally could be duplicated.
For me it's only happening on Chrome, Firefox work good

[brendan-donegan]

Also seeing this:

Node: 8.11.0
Protractor: 5.3.1
selenium-webdriver: 2.53.43

[kapilchokhawala]

are you guys sure is this related to Protractor, I am seeing it in WebdriverIO too?

[charlieargue]

Yup, same here, very intermittent though. Using async/await, but couldn't get .each() or .map() to work at all really, having some luck with .get(#).... like so (still get the same error, but less):

let length:number = await pageTracker.amountButtons.count();
    for(var i =0; i<length; i++) {
      await pageTracker.amountButtons.get(i).click();
    }

Because couldn't just do (like I would like to), because of the above error all the time:

pageTracker.amountButtons.each(async (amountBtn) => await amountBtn.click());

Node: 8.11.1
Protractor: 5.3.3
chromedriver_2.39
selenium-server-standalone-3.12.0.jar

[janjad]

I'm also getting the error, but at least workaround with lower selenium server version works for me...
See SeleniumHQ/selenium#5611 (comment)