443040 - qm-win2k3-pgo01 is failing reftests for bug 199692

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[Lukas Blakk [:lsblakk] use ?needinfo]

The test cases for this bug are being extremely noisy on qm-win2k3-pgo01:

*** 12174 ERROR FAIL | Element from doubly nested iframe returned is from
calling document | got [object HTMLHtmlElement], expected [object
HTMLIFrameElement] | /tests/content/html/document/test/test_bug199692.html
*** 12172 ERROR FAIL | Hit testing should bypass hidden elements. | got [object
HTMLIFrameElement], expected [object HTMLDivElement] |
/tests/content/html/document/test/test_bug199692.html
*** 12166 ERROR FAIL | (327,313) IDs should match (upper left corner of
textbox) | got [object HTMLIFrameElement], expected [object HTMLInputElement] |
/tests/content/html/document/test/test_bug199692.html

For example.

There are almost 70 lines of errors like this from yesterdays builds and
already 21 today.

Comment 1

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Of course now I realize I have no idea what the problem is and it's going to be hard to tackle.

Lukas, are you able to temporarily take over the box and reproduce the failures by running that mochitest on its own? Or by copying the build off the box and running the build locally?

Comment 2

[Lukas Blakk [:lsblakk] use ?needinfo]

I can try to grab the next build when it is done.  Each build clobbers the previous so all we have is the logs - but I can copy the next build for sampling and try running it locally.

Comment 3

[John O'Duinn [:joduinn] (please use "needinfo?" flag)]

Roc:

Is this something you (or whomever owns this test) can investigate? Lukas is trying to chase down intermittent problems on a whole bunch of unittests, so wont have time to do all these local steps to reproduce for all of these intermittent unittests.

If it helps, we could send you the objdir?

Also, what info should be added to the logs to make this kinda debugging easier in future?

Comment 4

[Robert O'Callahan (:roc) (email my personal email if necessary)]

If the build is published somewhere, that will help.

But sometimes we can't reproduce the problem using the same build because it's a problem specific to the test machine. In that case we have no choice but to call on people who have access to that machine to help.

Well, unless bug 438954 is fixed, which is the ultimate solution to all of this.

Comment 5

[Lukas Blakk [:lsblakk] use ?needinfo]

This has not been an issue since the move - resolving.

Comment 6

[Bernd]

this is now happening on tinderbox for 3 builds in a row

Comment 7

[Serge Gautherie (:sgautherie)]

Main failing box is
WINNT 5.2 mozilla-central qm-win2k3-moz2-01 dep unit test
but
WINNT 5.2 mozilla-central qm-win2k3-03 dep unit test
had it too.

Comment 8

[Ben Karel [eschew]]

The latest tinderbox results for qm-win2k3-moz2-01 dep unit test seems to indicate all tests for 199692 passing. I clobbered my mc build and couldn't reproduce any test failures.

If I can get binaries for a failing build, I can try to see what's going wrong, but I can't reproduce the failures locally, so I'm not sure what can be done...

Comment 9

[Lukas Blakk [:lsblakk] use ?needinfo]

This bug was for qm-win2k3-pgo01 which is now fx-win32-1.9-slave09 (pgo01), and is no longer failing this reftest.  If there are issues with reftest 199692 on other machines, please open a new bug.

Closing.