This started failing on Windows 8 both 32 and 64 bit versions
Error: Disconnect Error: Application unexpectedly closed
Mozmill version: 1.5.18
Firefox branch: default
Report link 64b: http://mozmill-ci.blargon7.com/#/functional/report/671677a5d9d5ca25f3cf5ae1c46e23c4
Report link 32b: http://mozmill-ci.blargon7.com/#/functional/report/671677a5d9d5ca25f3cf5ae1c46e29cb
Lets see if other branches are also affected.
Beta also seems to be affected
It is only happening on Windows 8.
So what's actually happens here is that we are stuck at "Transferring data from localhost..." in the add-on bar. This could be an issue with httpd.js on Windows 8 32/64. Jeff are you aware of anything like that?
Vlad, this is a critical failure so I want to see it at least fully investigated later today.
IT is trying to set me up a Windows 8 VM on the test box. Stay tuned!
Well I have the win 8 env now but its so strange the disconnect is not reproducible for me.
I'll have to dig in deeper, probably trigger some jobs again in jenkins.
Andreea or Alex, can one of you please take care of it ASAP until Vlad is back?
So I nailed this down even a bit more. The problem on that bug is the following line in the testcase file:
> window.open('popup_target.html', 'Test Popups ' + i, 'height=300,width=300');
Once it gets called loading of the page stops. It also happens with mozqa.com so it's definitely not a httpd.js issue.
Not sure why we end up with a disconnect here. It should at least run out of time in waitForPageLoad().
Loading the page outside of Mozmill works. So I would say it's an issue with Mozmill. I will file the bug once I found the real details.
So the underlying issue here is actually a bug in Firefox on Windows 8. I will file a bug soon.
The question why we do not run into a timeout for waitForPageLoaded() is still something I have to investigate and to fix in Mozmill.
There is nothing we can do here. So I will pull this from the task list and replace it by another bug.
Created attachment 662138 [details] [diff] [review]
Skipping the test for Windows. It applies cleanly to aurora, beta and release too.
Comment on attachment 662138 [details] [diff] [review]
Fixes have been landed on beta and aurora branches. For nightly it will follow soon. So I don't see a reason why we have to skip those patches now.
With the fix on bug 789773 landed on all the affected branches, the failure is gone now. Yay! We have a clean testrun on Windows 8 now.
I'm not sure why but it appears this has come back for Firefox 25.0a2:
(In reply to Anthony Hughes, Mozilla QA (:ashughes) from comment #15)
> I'm not sure why but it appears this has come back for Firefox 25.0a2:
I think this is more than likely bug 898194
That's correct, the recent failures were fallouts from bug 898194