Protractor hangs infinitelyon Chrome w/ shardTestFiles and maxInstances caps

The below piece of the config file is for parallel run, though it hangs if running one test at a time

the following error quite often accompanies

- Error: Timeout - Async callback was not invoked within timeout specified by jasmine.DEFAULT_TIMEOUT_INTERVAL.

• Protractor Version: 5.4
• Browser(s): Google Chrome 69.0.3497.100
• Operating System and Version Ubuntu

Chrome args:

        '--disable-infobars',
        '--disable-notifications',
        '--window-size=1440,900',
        '--headless',
        '--no-sandbox',
        '--disable-gpu',
        '--enable-logging',
        'v=1'

caps:

    capabilities: {
        browserName: 'chrome',
        chromeOptions: chromOpt,
        loggingPrefs: {
            'browser': 'ALL', // "OFF", "SEVERE", "WARNING", "INFO", "CONFIG", "FINE", "FINER", "FINEST", "ALL"
            'server': 'ALL',
            'driver': 'ALL'
        },

        shardTestFiles: true,
        maxInstances: 3
    },

Log:

17:01:44 
17:01:44 [15:01:44] I/testLogger - 
17:01:44 
17:01:44 [15:01:44] I/launcher - 2 instance(s) of WebDriver still running
17:31:06 Build timed out (after 90 minutes). Marking the build as aborted.
17:33:07 [15:33:07] I/testLogger - 

17:33:07 [15:33:07] I/testLogger - 
17:33:07 
17:33:07 [15:33:07] I/launcher - 1 instance(s) of WebDriver still running
17:33:07 /tmp/jenkins8334328386245671923.sh: line 18: 24330 Terminated              protractor test.conf.js --suite=regression --disableChecks
17:33:07 Build was aborted
17:33:07 [htmlpublisher] Archiving HTML reports...
17:33:07 [htmlpublisher] Archiving at PROJECT level /var/lib/jenkins/workspace/STAGING e2e/reports to /var/lib/jenkins/jobs/STAGING e2e/htmlreports/HTML_20Report

Before protractor launch export DBUS_SESSION_BUS_ADDRESS=/dev/null

From what I have read, the root cause of this issue might be chrome browser However IMHO, protractor should not hang but time out if this occurs

[aj-dev]

If you’re running tests on CI server in a docker container, try adding --disable-dev-shm-usage to Chrome args. It solved “Chrome not reachable”, “no such session” and related timeout issues for us when running sharded tests on multiple instances.

Looks like Chrome’s memory usage has changed drastically since v69

[Izhaki]

Same issue. Commenting onComplete solved it for us (shardTestFiles and maxInstances no commented).