Closed Bug 1032073 Opened 10 years ago Closed 7 years ago

Test failure 'aModule.tabBrowser is undefined' in /testAddons/testSearchAddons.js and /testGeolocation/testShareLocation.js

Categories

(Mozilla QA Graveyard :: Mozmill Tests, defect, P1)

defect

Tracking

(firefox35 disabled, firefox36 disabled, firefox37 disabled, firefox38 disabled, firefox39 disabled)

RESOLVED INVALID
Tracking Status
firefox35 --- disabled
firefox36 --- disabled
firefox37 --- disabled
firefox38 --- disabled
firefox39 --- disabled

People

(Reporter: andrei, Unassigned)

References

()

Details

(Whiteboard: [mozmill-test-failure][mozmill-test-skipped])

Attachments

(1 file)

Module:    teardownModule
Test:      /testAddons/testSearchAddons.js    
Failure:   aModule.tabBrowser is undefined
Branches:  mozilla-aurora
Platforms: OSX
Report:    http://mozmill-daily.blargon7.com/#/remote/report/59ac3f70f127c02da3a59eb1fc195f2e

Failed twice. Both on OSX.

Weird failure since `aModule.tabBrowser` is created in setupModule and used in the actual test. The failure happened in `teardownModule`...

There's a weird message in the Console _before_ the testrun started. I can't see how that could affect this failure, but here it is:
> 09:50:06 Started by user anonymous
> 09:50:06 ln builds/lastSuccessfulBuild /data/mozmill-ci/jenkins-master/jobs/mozilla-aurora_remote/lastSuccessful failed
> 09:50:06 java.nio.file.FileAlreadyExistsException: /data/mozmill-ci/jenkins-master/jobs/mozilla-aurora_remote/lastSuccessful
> 09:50:06  at sun.nio.fs.UnixException.translateToIOException(UnixException.java:88)
> 09:50:06  at sun.nio.fs.UnixException.rethrowAsIOException(UnixException.java:102)
> 09:50:06  at sun.nio.fs.UnixException.rethrowAsIOException(UnixException.java:107)
> 09:50:06  at sun.nio.fs.UnixFileSystemProvider.createSymbolicLink(UnixFileSystemProvider.java:456)
> 09:50:06  at java.nio.file.Files.createSymbolicLink(Files.java:996)
> 09:50:06  at sun.reflect.GeneratedMethodAccessor221.invoke(Unknown Source)
> 09:50:06  at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
> 09:50:06  at java.lang.reflect.Method.invoke(Method.java:606)
> 09:50:06  at hudson.Util.createSymlinkJava7(Util.java:1148)
> 09:50:06  at hudson.Util.createSymlink(Util.java:1064)
> 09:50:06  at hudson.model.Run.createSymlink(Run.java:1790)
> 09:50:06  at hudson.model.Run.updateSymlinks(Run.java:1771)
> 09:50:06  at hudson.model.Run.execute(Run.java:1682)
> 09:50:06  at hudson.model.FreeStyleBuild.run(FreeStyleBuild.java:43)
> 09:50:06  at hudson.model.ResourceController.execute(ResourceController.java:88)
> 09:50:06  at hudson.model.Executor.run(Executor.java:231)
> 09:50:06 [EnvInject] - Loading node environment variables.
> 09:50:06 Building remotely on mm-osx-108-3 (mac 64bit 10.8) in workspace jenkins/workspace/mozilla-aurora_remote
> 09:50:06 
> 09:50:08 Deleting project workspace... done
> 09:50:08 
> 09:50:08 Restoring workspace from build #89622 of project get_mozmill-tests
This hasn't failed again since the initial report (3 times total on the 29th and 30th June).
http://mozmill-daily.blargon7.com/#/remote/failure?app=Firefox&branch=All&platform=All&from=2014-06-29&test=%2FtestAddons%2FtestSearchAddons.js&func=teardownModule

But we have another instance, failked twice in /testGeolocation/testShareLocation.js:
http://mozmill-daily.blargon7.com/#/functional/failure?app=Firefox&branch=All&platform=All&from=2014-08-15&test=%2FtestGeolocation%2FtestShareLocation.js&func=teardownModule


Report: http://mozmill-daily.blargon7.com/#/functional/report/2561af7d8c83a450772e8c43801d39ee
Summary: Test failure 'aModule.tabBrowser is undefined' in /testAddons/testSearchAddons.js → Test failure 'aModule.tabBrowser is undefined' in /testAddons/testSearchAddons.js and /testGeolocation/testShareLocation.js
Happened again with Aurora on mm-ub-1310-64-1:
http://mozmill-daily.blargon7.com/#/functional/report/2561af7d8c83a450772e8c438087abd7

From the log I think it started from the previous test, testGeolocation started with teardown directly:

03:41:27 TEST-START | testGeolocation/testNotNowShareLocation.js | setupTest
03:41:27 TEST-START | testGeolocation/testNotNowShareLocation.js | testNotNowShareLocation
03:41:28 TEST-PASS | testGeolocation/testNotNowShareLocation.js | testNotNowShareLocation
03:41:28 TEST-START | testGeolocation/testNotNowShareLocation.js | teardownTest
03:41:28 TEST-END | testGeolocation/testNotNowShareLocation.js | finished in 960ms
03:41:28 *** WIFI GEO: shutdown called
03:41:29 
03:41:29 (process:29283): GLib-CRITICAL **: g_slice_set_config: assertion 'sys_page_size == 0' failed
03:41:33 TEST-START | testGeolocation/testShareLocation.js | teardownModule
03:41:33 ERROR | Test Failure | {
03:41:33   "exception": {
03:41:33     "message": "aModule.tabBrowser is undefined", 
03:41:33     "lineNumber": 34, 
03:41:33     "name": "TypeError", 
03:41:33     "fileName": "resource://mozmill/modules/frame.js -> file:///home/mozauto/jenkins/workspace/mozilla-aurora_functional/data/mozmill-tests/firefox/tests/functional/testGeolocation/testShareLocation.js"
03:41:33   }
03:41:33 }
03:41:33 TEST-UNEXPECTED-FAIL | testGeolocation/testShareLocation.js | teardownModule
Well, we do not list setupModule and teardownModule as executed as long as they pass. So setupModule could still have been run. It's very suspicious, and hard to tell anything as long as someone cannot reproduce it locally. :/
Same failure in another test /testGeolocation/testShareLocation.js, might be the same underlying issue:
http://mozmill-release.blargon7.com/#/functional/report/2f56cb3a3728c2c47cd1b44f77504a8d
And failed again, 
http://mozmill-release.blargon7.com/#/functional/report/2f982f72826307fed840a3b11cd6f5f6

Same as before:
> 22:55:53 TEST-START | testGeolocation/testNotNowShareLocation.js | setupTest
> 22:55:53 TEST-START | testGeolocation/testNotNowShareLocation.js | testNotNowShareLocation
> 22:55:54 TEST-PASS | testGeolocation/testNotNowShareLocation.js | testNotNowShareLocation
> 22:55:54 TEST-START | testGeolocation/testNotNowShareLocation.js | teardownTest
> 22:55:54 TEST-END | testGeolocation/testNotNowShareLocation.js | finished in 1009ms
> 22:55:55 
> 22:55:55 (process:15209): GLib-CRITICAL **: g_slice_set_config: assertion 'sys_page_size == 0' failed
> 22:55:59 TEST-START | testGeolocation/testShareLocation.js | teardownModule
> 22:55:59 ERROR | Test Failure | {
> 22:55:59   "exception": {
> 22:55:59     "message": "aModule.tabBrowser is undefined", 
> 22:55:59     "lineNumber": 34, 
> 22:55:59     "name": "TypeError", 
> 22:55:59     "fileName": "resource://mozmill/modules/frame.js -> file:///home/mozauto/jenkins/workspace/release-mozilla-beta_functional/data/mozmill-tests/firefox/tests/functional/testGeolocation/testShareLocation.js"
> 22:55:59   }
> 22:55:59 }
> 22:55:59 TEST-UNEXPECTED-FAIL | testGeolocation/testShareLocation.js | teardownModule
> 22:55:59 TEST-END | testGeolocation/testShareLocation.js | finished in 6ms
Please follow the order of started methods here. This is a restart test and teardownModule should not have been executed before `testNotNowShareLocationPersisted` has been run. What I think happened here is that after the restart in teardownTest, Firefox came up again, the test method was not found, and no setupTest has been executed. Surprisingly I can see this in my refactored software update tests.
This did not happen again, for testSearchAddons there are 3 reports in june when the bug was filed. For testShareLocation I don't see anything else than the one Andrei mentioned. I think this was more related to bug 972912, as I don't see setupModule to have run.
Depends on: 972912
Lets skip this on Linux, seems the only one affected for testShareLocation
So, because there were many failures lately on beta & ubuntu, we decided we should skip testShareLocation.js on Linux for mozilla-beta branch. Here is the skip patch.
Attachment #8549507 - Flags: review?(andreea.matei)
Attachment #8549507 - Flags: checkin?(andreea.matei)
Comment on attachment 8549507 [details] [diff] [review]
skiptstshrlctn.patch

Review of attachment 8549507 [details] [diff] [review]:
-----------------------------------------------------------------

http://hg.mozilla.org/qa/mozmill-tests/rev/00f0b67f6e3d (beta)
Attachment #8549507 - Flags: review?(andreea.matei)
Attachment #8549507 - Flags: review+
Attachment #8549507 - Flags: checkin?(andreea.matei)
Attachment #8549507 - Flags: checkin+
Whiteboard: [mozmill-test-failure] → [mozmill-test-failure][mozmill-test-skipped]
In this case please really really work on a reproduction case. This would be so important to get this fixed if it fails a lot.
Priority: P4 → P1
Transplanted to release too, we had failures with today's 35.0.1:
https://hg.mozilla.org/qa/mozmill-tests/rev/d94d5580c40e (release)
The skip for testShareLocation.js should have been across all OS. It's also failing a lot on OS X now. I will have a skip patch shortly.
Mozmill tests have been superseded by Marionette tests.
Status: NEW → RESOLVED
Closed: 7 years ago
Resolution: --- → INVALID
Product: Mozilla QA → Mozilla QA Graveyard
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: