Closed Bug 1148775 Opened 9 years ago Closed 9 years ago

Intermittent stream of updater xpcshell failures starting with unit_service_updater/bootstrapSvc.js | xpcshell return code: 0

Categories

(Toolkit :: Application Update, defect)

Product:
Toolkit
Component:
Application Update
Platform:
x86
Windows 8
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

()

Status:
RESOLVED WORKSFORME

People

(Reporter: philor, Unassigned)

References

Details

(Keywords: intermittent-failure) 

Not sure what we've been doing with these, ignoring them, maybe misstarring some as bug 1118175 like bug 1118175 comment 46.
08:10:03     INFO -  TEST-START | toolkit/mozapps/update/tests/unit_service_updater/bootstrapSvc.js
08:10:03  WARNING -  TEST-UNEXPECTED-FAIL | toolkit/mozapps/update/tests/unit_service_updater/bootstrapSvc.js | xpcshell return code: 0
08:10:03     INFO -  TEST-INFO took 298ms
08:10:03     INFO -  >>>>>>>
08:10:03     INFO -  (xpcshell/head.js) | test MAIN run_test pending (1)
08:10:03     INFO -  "08:10:03:974 | TEST-INFO | ../data/xpcshellUtilsAUS.js | [attemptServiceInstall : 1904] unable to copy new maintenance service into the maintenance service directory: C:\\Program Files (x86)\\Mozilla Maintenance Service, Exception: [Exception... "Component returned failure code: 0x80520015 (NS_ERROR_FILE_ACCESS_DENIED) [nsIFile.moveTo]"  nsresult: "0x80520015 (NS_ERROR_FILE_ACCESS_DENIED)"  location: "JS frame :: ../data/xpcshellUtilsAUS.js :: attemptServiceInstall :: line 1892"  data: no]"
08:10:03     INFO -  The account running the tests on the build systems should have write access to the maintenance service directory!
08:10:03     INFO -  ../data/xpcshellUtilsAUS.js:attemptServiceInstall:1907
08:10:03     INFO -  ../data/xpcshellUtilsAUS.js:shouldRunServiceTest:1643
08:10:03     INFO -  C:/slave/test/build/tests/xpcshell/tests/toolkit/mozapps/update/tests/unit_service_updater/bootstrapSvc.js:run_test:8
08:10:04     INFO -  C:\slave\test\build\tests\xpcshell\head.js:_execute_test:504
08:10:04     INFO -  -e:null:1
08:10:04     INFO -  null:null:0
08:10:04     INFO -  exiting test

The test throws because "The account running the tests on the build systems should have write access to the maintenance service directory!"

Per bug 1067756 comment #23 the systems should have write access.

phil, for the change made in bug 1067756 comment #23 to take affect the system would have needed to be rebooted. I can't imaging that system hadn't been rebooted but is there a way to verify that it has?

Q, can you check that there wasn't a problem with the GPO?
Flags: needinfo?(q)
Flags: needinfo?(philringnalda)
phil, would it be easier to identify if the failure was a test failure vs. throwing?
Well, "foo.js | the particular thing that failed" is certainly always better than the standard xpcshell "foo.js | xpcshell return code: 0", but we don't actually have a significant number of intermittent update tests right now, so now that it's filed I think the only thing that will get in the way of them being starred correctly is the broken way the harness fails to kill processes from previous timeouts and leaves a litter of "WindowsError: [Error 5] Access is denied" above you.

But given that https://secure.pub.build.mozilla.org/buildapi/recent/t-w864-ix-146?numbuilds=500 plus a search for xpcshell says that this one slave has failed a dozen trunk runs since it got broken on the 20th, it's possible that it's the only one doing it, and all we'll have to worry about it persuading it to behave.
Flags: needinfo?(philringnalda)
Blocks: t-w864-ix-146
I wiil take a look at the host and I will run an audit acroos the nodes to check permissions
Flags: needinfo?(q)
I've updated the test to display a clearer error without the error string that confused this with the other bug. Closing this as wfm since this is a build system issue with another bug to fix it and all has been well since the slave has been taken out.
Status: NEW → RESOLVED
Closed: 9 years ago
Resolution: --- → WORKSFORME
Comment #9 is not the same failure at all... just look at the previous comments with logs.

failed: 26 is SERVICE_UPDATER_SIGN_ERROR and indicates either a signing error for the build or a builder issue.
It was actually a broken test slave, which I had already disabled. Sometimes, it's just hard to resist the lure of a suggested bug.
You need to log in before you can comment on or make changes to this bug.