Open Bug 1336760 Opened 5 years ago Updated 3 years ago

Intermittent TEST-UNEXPECTED-TIMEOUT | testing/web-platform/tests/payment-request/allowpaymentrequest/allowpaymentrequest-attribute-same-origin-bc-containers.https.html | expected OK

Categories

(Core :: DOM: Core & HTML, defect, P5)

defect

Tracking

()

People

(Reporter: intermittent-bug-filer, Unassigned)

References

(Blocks 1 open bug)

Details

(Keywords: intermittent-failure, leave-open, Whiteboard: [stockwell disabled])

Attachments

(1 file)

See Also: → 1336775
Intermittently failing since added in bug 1335397.
Blocks: 1335397
keeping an eye on this, the failure rate is steady, if we have more pushes or retriggers this will cross our 30 failures/week threshold.  oddly this is primarily failing on pgo and in e10s mode.
Component: web-platform-tests → DOM
Product: Testing → Core
Version: Version 3 → unspecified
Whiteboard: [stockwell needswork]
:jgraham, as there is no bugzilla component associated with this, can you help find someone to look at this?  Between this bug and bug 1336775 there are 128 failures in the last 7 days- I suggest disabling this temporarily until we can get a fix for this.
Flags: needinfo?(james)
looking into this the test is marked expected fail on all platforms- and we are hitting a failure, just a timeout after 10 seconds.


all on opt/pgo, not on debug, so odd that we hit this- maybe there is a timing issue?

oddly I am unable to find this test in github:
https://github.com/w3c/web-platform-tests/search?utf8=%E2%9C%93&q=attribute-same-origin-bc-containers&type=


but browsing the repo I find:
https://github.com/w3c/web-platform-tests/blob/master/payment-request/allowpaymentrequest/allowpaymentrequest-attribute-cross-origin-bc-containers.https.html

and it looks like these typically fail on timeout somewhere in the common.sub.js file:
https://github.com/w3c/web-platform-tests/blob/master/payment-request/allowpaymentrequest/common.sub.js

all of these tests have been updated 2 months ago in github, started failing Feb 5th (about when we would have added these to the mozilla trees).  Around March 23rd or 24th this picked up frequency, more instances on mozilla-aurora which makes this hard to figure out- March 25th we see more win8/osx failures than win7- maybe that is a clue.

We landed 4 content processes on March 17th, maybe this falls into alignment with that.

Here is an example failure log:
https://treeherder.mozilla.org/logviewer.html#?repo=autoland&job_id=88193698&lineNumber=4821


Simon, would you have ideas of what is going on with these tests and how they timeout?  any ideas for debugging?
Flags: needinfo?(james) → needinfo?(zcorpan)
I guess there are several things that could cause a timeout. If echo-PaymentRequest.html fails to load in <frame> or <iframe>, it would time out since those elements don't fire an 'error' event. If it loads but somehow fails to receive or send the message (window.postMessage), it would time out. If it gets a message where the data is null or undefined, line 10 would throw and would not be caught by the harness since it's not t.step-ed of any test (maybe it could have a separate "setup" test for this). That's about what I can think of right now, but possibly could be something else entirely.
Flags: needinfo?(zcorpan)
:gbrown, another test that I could use some help debugging?  Possibly there is some quick debugging to fix this?  I might get to this later this week- but do let me know if you have time to work on this.
Flags: needinfo?(gbrown)
to top this off, we expect all tests here to fail, so this test is not providing a lot of value.  This is because firefox doesn't support paymentRequest (requested to implement via bug 1318987)

Overall this test repeats 4 times, one each for: iframe, frame, object, embed;  we seem to always finish iframe, frame, but timeout on object.

As there are no screenshots, I am unclear if we have loaded anything, so we do not know if we loaded the data or are missing.

While there is a real problem here, with the tests timing out, there is no developer stepping up to fix the issues, so I will just turn off the tests.  In addition we have bug 1336775 failing at the same frequency.
Flags: needinfo?(gbrown)
Attachment #8857082 - Flags: review?(gbrown) → review+
Pushed by jmaher@mozilla.com:
https://hg.mozilla.org/integration/mozilla-inbound/rev/ffb859addb94
disable paymentrequest tests as firefox doesn't support this yet. r=gbrown
Keywords: leave-open
Whiteboard: [stockwell needswork] → [stockwell disabled]
Summary: Intermittent TEST-UNEXPECTED-TIMEOUT | /payment-request/allowpaymentrequest/allowpaymentrequest-attribute-same-origin-bc-containers.https.html | expected OK → Intermittent TEST-UNEXPECTED-TIMEOUT | testing/web-platform/tests/payment-request/allowpaymentrequest/allowpaymentrequest-attribute-same-origin-bc-containers.https.html | expected OK
Blocks: 1318987
No longer blocks: 1318987
Priority: -- → P5
Component: DOM → DOM: Core & HTML
You need to log in before you can comment on or make changes to this bug.