Zombie Process Left Running

[ljohnston]

Create the following task config:

protractor_webdriver: {
    start: {
    }
},

Then simply run grunt protractor_webdriver:start. After grunt exits a java process will still be running on port 4444. When protractor_webdriver:start is called as part of my integration test task which then runs the tests, however, after grunt exits there is no lingering zombie process. Note that running with debug does claim that the server gets shutdown:

$ grunt protractor_webdriver:start
...
Starting Selenium server
Started Selenium server: http://127.0.0.1:4444
Shutting down Selenium server: http://127.0.0.1:4444

This is with version 0.2.0. Rolling back to 0.1.9 fixes the issue.

[BrodaNoel]

+1 I have the same issue! I promiss you invite a coffee cup if you solve this in less than 24 hours :dash:

[BrodaNoel]

I'm rolling back to 0.1.9 and still broken. You could create a protractor_webdriver:stop by default and we can call it at the end. My 2 cents.

[BrodaNoel]

I had a lot of jasmine-reportes added, Console, screenshot, and more. When I remove all them, the fix disapear. So, the issue is related to the browser.takeScreenshot() or something like that.

[BrodaNoel]

Hi Steffen Eckardt! Look... I just install the plugin, and when I run the tests, look at what's append. Here 2 screenshots.

[BrodaNoel]

Ooook, some news about it:

Seems that the http.get(shutDown) never is executed. So, what I did was add this sub-task: shell:killPlease Whit that options:

killPlease: {
                options: {
                    stdout: true
                },
                command: 'curl http://127.0.0.1:4444/selenium-server/driver/?cmd=shutDownSeleniumServer'
            }

And now, I get the prompt again, but, why is it necessary? Good luck!

[jeandat]

Same issue here. Grunt output says server is shutdown but that's not the case.