*** 4196 ERROR FAIL | testFormElementObserver | 5assertions, 1 failures, 0 errors Failure: assertEqual:expected "1", actual "0" This test is in: /dom/tests/mochitest/ajax/prototype/test/unit/form.html The test seems to be using some observers to look for changes to a form field value, but it's not really clear to me how this test works in the first place. It looks like it has race conditions? Checkins in range prior to the build exhibiting this error: http://bonsai.mozilla.org/cvsquery.cgi?module=PhoenixTinderbox&date=explicit&mindate=1186589653&maxdate=1186594704
I ran mochitest on a tree I pulled ~5pm today, and the test completed without any failures.
I landed jresig's patch in bug 375543 in an effort to get a better stack.
chrome://mochikit/content/browser/browser/fuel/test/browser_Application.js NEXT ERROR FAIL - Exception thrown - ReferenceError: Application is not defined chrome://mochikit/content/browser/browser/fuel/test/browser_ApplicationPrefs.js NEXT ERROR FAIL - Exception thrown - ReferenceError: Application is not defined chrome://mochikit/content/browser/browser/fuel/test/browser_ApplicationStorage.js NEXT ERROR FAIL - Exception thrown - ReferenceError: Application is not defined chrome://mochikit/content/browser/browser/fuel/test/browser_Bookmarks.js FAIL - Exception thrown - ReferenceError: Application is not defined chrome://mochikit/content/browser/browser/fuel/test/browser_Browser.js FAIL - Exception thrown - ReferenceError: Application is not defined chrome://mochikit/content/browser/browser/fuel/test/browser_Extensions.js FAIL - Exception thrown - ReferenceError: Application is not defined http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1186635938.22213.gz&fulltext=1
The "random" browser failures in comment 3 seem to have disappeared, fwiw.
So... Who's looking at what happened here? mxr shows that sayrer removed test_largemenu.xul from /toolkit/content/tests/chrome/Makefile.in at 2am for "bustage fix", which didn't seem to help (should it be put back in?). The ~10:30am build has a note saying that rcampbell removed forms.html to force a build. The next build failed with: *** 4194 INFO PASS | testSize *** 4195 ERROR FAIL | Test timed out. | *** 4197 INFO Running /tests/dom/tests/mochitest/ajax/scriptaculous/test_Scriptaculous.html... Comparing to the previous log, it looks like the forms.html test is now completely broken, and times out before logging anything at all.
qm-xserve01 seems to have gone green this afternoon; did the hardware problem get fixed, or something else fix this? Or just random success?
I really don't know what's going on. We had some green runs on the machine this afternoon and then a new failure appeared. *** 8463 INFO PASS | testFormElementEventObserver *** 8464 ERROR FAIL | testFormElementObserver | 5 assertions, 1 failures, 0 errors Failure: assertEqual: expected "1", actual "0" Line #180 We're investigating the hardware problem that is causing the machine to slow down. In the meantime, I think test results from this machine should be considered untrustworthy.
Also, keep in mind that the slowness of the machine is causing severe lag with respect to build and result times. A test run that completes may be operating on checkins from over an hour ago. Please use bm-xserve07 for accurate results until we get this fixed.